Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript. Biodiversity loss threatens ecosystem services on which humanity depends, yet the strength, form and context-dependency of biodiversity–ecosystem service and biodiversity–ecosystem function relationships remain poorly quantified. Here we synthesize 423 studies to determine how biodiversity affects ecosystem functioning and service delivery. We show that regulating services, including water quality and pollination, often exhibit linear relationships, suggesting an additive role for diversity. While biodiversity enhances most functions and services, our effect-size analysis shows that oceanic ‘blue carbon’ sequestration is exceptionally sensitive, revealing how critical marine biodiversity is for climate mitigation. Conversely, services such as hazard regulation appear relatively insensitive to biodiversity, instead depending on a few functionally unique foundational species whose persistence nonetheless is interdependent on the wider community. Our new synthesis enables the forecasting of impacts, for example, that biodiversity loss will reduce pest regulation in countries experiencing population growth and intensive development, and it also highlights how ecosystem services might be used to anticipate future problems. Similar content being viewed by others Food web complexity underlies biodiversity effects on ecosystem functioning Biodiversity offsets, their effectiveness and their role in a nature positive future Resistance of ecosystem services to global change weakened by increasing number of environmental stressors Explore related subjects Climate-change impacts Ecological modelling Ecosystem Services Assessment and Land Use Dynamics Ecological research highlights the close relationship between biodiversity and a wide range of critical ecosystem services and ecosystem functions, from pollination to carbon sequestration 1 , 2 , 3 , 4 . Anthropogenic pressures (for example, land-use change and introduction of exotic species) have driven an unprecedented number of species, estimated at 1 million, towards extinction, threatening ecosystem services 4 , 5 , 6 , 7 , 8 . While global policies aim to halt this decline 9 , 10 , effective action requires a quantitative and synthetic understanding of how biodiversity influences ecosystem functions and services. Here, we follow the ecosystem service classification framework of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services 11 (IPBES), which recognizes that ecosystem functions (processes that occur within an ecosystem) often underpin or constitute ecosystem services when they contribute to human well-being. For example, primary production can be an ecosystem service when it supports food provision or climate regulation. Accordingly, we classify response variables as ecosystem services where they align with IPBES nature’s contributions to people categories or ecosystem functions, and as ecosystem functions when they do not always provide direct or indirect benefits to people, to ensure comprehensive coverage of key ecosystem processes. Many functions included here may also be described as ecosystem services in other contexts 12 , 13 , 14 , reflecting overlapping use of the terms in literature. We also chose to explicitly separate carbon sequestration from other services and to assess its across terrestrial and marine realms, given the interest in biodiversity as a climate-regulator 15 . While the consequence of biodiversity declines on ecosystem functioning is well-established, the functional form and scalability of biodiversity–ecosystem service (BES) and biodiversity–ecosystem function (BEF) relationships remain a fundamental challenge. BEF and BES relationships are often classified simplistically as linear or saturating, with no clear consensus on the scalability of data to real-world patterns 1 , 16 . Whether these relationships are typically linear, implying all species including rare and/or pollution-sensitive taxa contribute equally, or saturating, suggesting widespread functional redundancy, is poorly resolved at a global scale 16 , 17 , 18 . For a limited set of processes, such as primary production in plant communities, mechanistic explanations of biodiversity effects, such as selection, complementarity or portfolio effects, are relatively well studied 19 , 20 , 21 . However, our understanding of the generality of the qualitative (linear versus saturating) and quantitative relationship between biodiversity and ecosystem services and functions is limited as it is unclear if these mechanisms apply to the full range of ecosystem services. Foundational BEF research has often focused on ecosystems that are particularly amenable to experimental manipulation, such as small-scale experimental plots in grasslands 18 , 22 . Although recent studies have expanded the scope of BEF research across a wider range of ecosystems and spatial scales, a comprehensive global synthesis (that is, across ecosystem types and regions) is still lacking 3 , 23 . Here we present a new global database of 423 studies of biodiversity relationships with ecosystem functioning and services, more than twice as large as the next largest database 24 , representing a major expansion in ecological, methodological and geographical diversity (Fig. 1 ). This database combines experimental and observational evidence across terrestrial, freshwater, marine and estuarine systems, enabling, for the first time, a consistent global assessment of how biodiversity measurements (species number, functional group and taxonomic group) influence multiple categories of ecosystem service. Using this framework, we evaluate the functional forms of BES/BEF relationships, quantify their sensitivity to biodiversity loss, identify generalized factors shaping these patterns and model future trajectories in pest regulation, which may contribute to food and feed provision under alternative population growth and development scenarios. a – d , Study sites used to analyse global BES and BEF trends ( a ), with studies colour-coded according to ecosystem type and inserts showing North America ( b ), Europe ( c ) and China ( d ). Study sites are predominantly terrestrial and located in developed countries. e , There are 1,759 datasets from terrestrial studies compared to 183 in freshwater, 153 in marine and 23 in estuarine systems. f , There are 20% more observational ( n = 1,079) datasets in our database compared with experimental ( n = 879) datasets. g , Grain size ranged from 0.0000007 m 2 to 260,018,758 m 2 with a median value of 1 m 2 across datasets. World map data from Natural Earth ( https://www.naturalearthdata.com/ ). Functional variation in BES and BEF relationships While positive BES and BEF relationships are commonly described 16 , 25 , the functional forms of biodiversity effects remain poorly quantified across and within ecosystem services categories 11 . Across 23 categories of BES and BEF relationships, quadratic ( n = 11) and linear ( n = 9) relationships were most prevalent, with only two best-described by a logarithmic model, suggesting that most ecosystem services continue to increase rather than saturate at low levels of biodiversity (Fig. 2 ). Dataset count values are shown in Fig. 3b . The relative likelihood that an ecosystem service category (rows; measured by Akaike weights with error bars representing ±1 standard error (s.e.)) is fit by linear and saturating (quadratic or logarithmic) models. A high average Akaike weight value indicates a greater likelihood that the model type best fits the data than other model types. Most BES and BEF relationships are best fit by quadratic (11/23) or linear (10/23) models, whereas only two are best fit by a logarithmic model (2/23). We note that these results are not induced by a splitting of variance among two saturating models; we find qualitatively identical results when comparing the best-fitting saturating model with linear (46% supporting linear and 54% supporting saturating). Data are sorted by the highest to the lowest mean value of the quadratic model. Response variables marked with an asterisk (*) are classified as ecosystem functions; all other response variables are ecosystem services 11 . Although saturating relationships are often reported 3 , 26 , 27 , 28 , we find that linear relationships are also common, with their prevalence varying among ecosystem services and ecosystem functions categories. For some functions, such as terrestrial net primary productivity (NPP), different models performed equally well, suggesting contextual factors (for example, region and study type) modulate the form of the relationship. Variation in BES and BEF relationships partly reflects methodological choices as biodiversity metric classes differed among ecosystem services (Cramér’s V = 0.34), and the form of relationships varied with metric type, with functional and phylogenetic diversity metrics more often producing saturating responses whereas taxonomic diversity evenly distributed across linear and saturating models (Extended Data Fig. 1 ). Importantly, explanatory power varied substantially among ecosystem service and function categories and among model forms. Most biodiversity–ecosystem relationships showed low-to-moderate r 2 values (for example, r 2 < 0.4; Supplementary Fig. 1 ), indicating that biodiversity is an important but context-dependent driver of ecosystem functioning and service delivery. However, explanatory power remained weak for some services, such as hazard regulation, implying stronger influence of environmental context or functionally unique species. Regulating services, including water quality, hazards and pollination, were typically linear, indicating a broad equivalent role for biodiversity (for example, species number, functional group or taxonomic group) in maintaining them. By contrast, oceanic carbon sequestration and learning and inspiration exhibited saturating relationships, indicating higher functional redundancy among species. While uneven data distribution, particularly uneven representation across ecosystem types and spatial grain (Fig. 1 ), and across ecosystem service and ecosystem function categories may influence these findings, the observed variation in BES and BEF relationships, including within well-studied categories, suggests that some services follow linear relationships while others tend towards saturation. Any changes in biodiversity are expected to affect ecosystem services and function when biodiversity levels fall below the saturation point. The prevalence of linear and saturating relationships suggests that some ecosystem services and functions peak only at the highest observed levels of biodiversity, underscoring the importance of the richest communities. However, linear trends may either reflect limited sampling of high-biodiversity systems or genuinely linear ecological responses in which linear relationships may be intrinsic to some ecosystem service and function types. No ecosystem service or function can rise indefinitely with biodiversity, but our data suggest saturation limits have rarely been reached within the observed biodiversity gradients in the empirical literature, implying functional redundancy may be overestimated. In other words, individual species contribute more uniquely to ecosystem functioning, and ecological uniqueness plays a key role in sustaining service delivery, consistent with the wide variation in saturation levels across ecosystems observed. However, in some systems, saturation is reached relatively quickly, such as decomposition in freshwater systems, which may saturate at low levels of biodiversity 29 For saturating relationships, the asymptote, where additional species no longer substantially increase ecosystem service or function, at a median of ~4 (± median absolute deviation of 2) species in freshwater, ~14 (±9) in terrestrial and ~23 (±17) in marine ecosystems (Extended Data Fig. 2 ). Faster saturation in freshwater systems may reflect their smaller scale and higher functional redundancy, particularly as they are characterized by relatively generalist food webs, whereas the vast scale and high trophic complexity and prevalence of mutualisms and co-evolutionary interactions in marine systems, such as coral reefs, may require higher biodiversity to sustain function 1 , 30 , 31 , 32 , 33 . Marine systems may also display low species multifunctionality, meaning that individual ecosystem services can often be supported by a small subset of species, but maintaining a wide range of ecosystem services requires a high overall diversity 33 . While our data show differences in median saturation values across ecosystem types, there was a wide range of saturation values, indicating that differences in experimental design, scale and habitat may also shape BES and BEF relationships (see the ‘Influence of spatial grain, ecosystem type and study type on BES and BEF trends’ for more details), warranting closer examination in future studies. Quantifying biodiversity effects on ecosystem services and ecosystem functions The variation in functional BES and BEF forms make it challenging to understand how ecosystem service and function categories respond to biodiversity. To overcome this, we use a second approach that allows us to explore relationships across forms. To our knowledge, this is the first analysis to quantify BES and BEF relationships across multiple functional forms using a standardized effect-size framework. We calculated the effect size, as the response of services and functions to increasing biodiversity (Fisher’s Z -transformed correlation coefficient), for each dataset ( Methods ; n = 1,959) 34 . Higher biodiversity enhanced most BES and BEF, from carbon capture to food production (Fig. 3 ). While the magnitude of this effect varied across ecosystem service and function categories (regulation of air quality, Fisher’s z score (Zr) −0.03, very weak; oceanic carbon sequestration, Zr 1.48, very strong), the dependence of critical regulatory services on biodiversity, until the point of saturation, is unequivocal. The imperative to foster and safeguard biodiversity is particularly acute for marine ecosystems. Oceanic carbon sequestration, which is a significant part of the global carbon cycle, exhibits by far the strongest positive response to biodiversity of any service or function category 35 . This strong sensitivity may reflect the strong biological control of marine carbon cycling, including phytoplankton productivity, microbial loop dynamics and the biological carbon pump, all of which are closely linked to community composition and trophic structure. However, this result should be interpreted cautiously, as oceanic carbon sequestration is relatively underrepresented ( n = 7), with far fewer datasets in our database than for terrestrial carbon sequestration ( n = 154), highlighting a critical gap in current knowledge (Fig. 3 ). This exceptional sensitivity implies that blue carbon sinks are both vulnerable to biodiversity loss and dependent on sustaining diverse communities 36 , 37 . Maximizing the efficacy and long-term viability of blue carbon strategies will require moving beyond habitat preservation towards explicitly incorporating biodiversity as a key component 38 , 39 . a , b , Comparison of biodiversity responses among ecosystem service and function categories, as effect size (Zr) ( a ) and the number of datasets used to calculate effect size ( b ). Effect size data are average ±1 standard error (s.e.), calculated on datasets within each ecosystem service category, where a high effect size value indicates a stronger effect of biodiversity on ecosystem service compared with a lower value. See Extended Data Fig. 4 for effect sizes split by study type (experiment versus observation). Higher biodiversity enhanced most ES, from carbon capture to food production, in particular oceanic and terrestrial carbon sequestration were strongly dependent on biodiversity. Conversely, some services, such as hazard regulation, appeared insensitive to biodiversity. Notably, these effects vary by habitat type, spatial grain and study type. Response variables marked with an asterisk (*) are classified as ecosystem functions; all other response variables are ecosystem services 11 . Conversely, some services, such as hazard regulation, exhibited weak overall sensitivity to biodiversity in the effect-size analysis, despite often showing linear relationships in the functional form analysis ( Zr = 0, r 2 = 0.17; Figs. 2 and 3 and Supplementary Fig. 1 ). This may reflect functional uniqueness, where service provision is dominated by a few foundational species or ecosystem engineers that define the ecosystem itself. For example, sand dunes are stabilized by one or two key shrub species that create a natural barrier against flooding and erosion 40 . Here, the service is provided by the foundational species, and the system’s direct performance in this role is not necessarily enhanced by the addition of other species. However, maintaining broader biodiversity may still be important for sustaining the conditions that allow these foundational species to persist and function effectively 38 , 39 , 41 , 42 , 43 . In systems governed by functionally unique species, maintaining ecosystem services and functions may rely more on the persistence of these taxa than on biodiversity per se. This approach aligns with the concept of ecosystem-specific ‘planetary boundaries,’ which are thresholds of control variables that, when crossed, can lead to the collapse of ecosystem functions 44 . Identifying a priori which systems are shaped by uniqueness versus redundancy is a critical challenge. Our data suggest that developing a typology of species and system traits is a promising avenue for future research to guide conservation priorities. Beyond further data synthesis, progress will depend on a new generation of targeted field experiments explicitly designed to disentangle the roles of uniqueness and redundancy in sustaining ecosystem functioning, thereby providing direct empirical tests of these theoretical predictions. Influence of spatial grain, ecosystem type and study type on BES and BEF trends Effect-size analyses offer robust insights into BES and BEF relationships, but they do not fully explain variation depending on ecosystems’ context. Contextual factors, such as spatial grain (that is, the spatial resolution at which biodiversity and ecosystem services are measured, from fine-scale plots to broader landscape units), ecosystem type and study type, may influence both the strength and form of biodiversity relationships with ecosystem services and functions. This is because ecological processes, including species interactions and environmental heterogeneity, operate differently across spatial scales, potentially amplifying or obscuring biodiversity effects. Understanding such patterns is essential to interpreting the generality of BES and BEF findings and improving forecasts under global change. Using a linear mixed-effect model, we tested for BES and BEF relationships with spatial grain, ecosystem type and study type as factors ( Methods and Supplementary Table 1 ). The influence of spatial grain on BES and BEF was highly context dependent. Across all ecosystem services and functions, biodiversity effects strengthened with increasing grain ( β = 210.26, P < 0.001; Supplementary Table 1 ), indicating that its contributions become more apparent at broader spatial scales. This pattern may partly reflect a beta-diversity effect, whereby spatial turnover in species composition enhances functional complementarity across sites, amplifying aggregate ecosystem service delivery at coarser grains, as predicted by spatial insurance and complementarity theories. However, responses varied substantially among ecosystem services and functions. For example, biodiversity contributions to regulation of hazards and extreme events, NPP and biomass turnover all declined significantly with increasing spatial grain (Supplementary Table 1 ). This scale-dependence is consistent with forest plot studies where positive BES and BEF relationships at fine grains (for example, 20 m 2 ) can become negative at coarser grains (for example, 50 m 2 ) 45 . Averaging over larger areas probably masks the fine-scale heterogeneity and species interactions that fuel ecosystem service and function provision 27 , 39 . Consequently, failing to account for measurement scale can fundamentally alter conclusions about ecosystem stability and responses to environmental change. Second, ecosystem type significantly influenced BES and BEF relationships and highlighted data gaps in literature. Although our literature review was not designed to quantify ecosystem type biases per se, the differences we found in sampling, particularly the sparse coverage of freshwater and marine systems, are unlikely to reflect sampling artefacts alone but rather systematic biases in research focus and accessibility, with terrestrial systems historically receiving disproportionate attention (Extended Data Fig. 3 and Supplementary Table 1 ). Overall, biodiversity effects were positive in freshwater systems ( β = 0.34, P = 0.01), negative in marine systems ( β = –0.25, P = 0.016) and not significant in terrestrial systems ( P > 0.05). However, these averages mask strong variation across ecosystem service categories. For example, biodiversity significantly enhanced biomass turnover rate in marine systems, reduced it in freshwater systems and had no effect in terrestrial systems. By contrast, for maintenance of options, biodiversity effects were consistently positive in both freshwater and marine environments. Such differences may reflect underlying ecological processes, such as higher functional redundancy, or may arise from sampling and design biases. These patterns should be interpreted in the context of uneven sampling across the global BES and BEF literature, with strong representation of terrestrial systems and fine spatial grains (for example, quadrats and individual plots), and limited coverage of freshwater and marine ecosystems and broader-scale studies (for example, landscapes and regions) (Fig. 1 and Supplementary Table 4 ). While overall biodiversity effects are robust, some ecosystem differences and functional forms may shift as additional data become available from undersampled systems and scales. Finally, observational studies revealed modest but statistically stronger biodiversity effects than experiments ( β = 0.03, P = 0.005), demonstrating that study type significantly shapes BES and BEF trends. This overall effect size is small relative to other predictors in our models and should be interpreted as highlighting service-specific contrasts rather than a universal shift per se. Biodiversity effects were pronounced in observational studies for multifunctionality ( β = 0.48), pollination and dispersal of seeds and other propagules ( β = 0.37) and NPP ( β = 0.19). Effects were weaker and negative for biomass turnover rate ( β = –0.03) and regulation of hazards and extreme events ( β = –0.13), with no detectable differences for several other services, including evapotranspiration and food and feed (Supplementary Table 1 and Extended Data Fig. 4 ). These findings indicate that experiments are critical for revealing mechanisms such as complementarity, overyielding and sampling effects but typically operate at small scales and simplified conditions potentially underestimating cumulative or context-dependent impacts 46 , 47 , 48 , 49 . Observational studies capture broader spatial and temporal scales, larger biodiversity gradients (including rare and functionally unique species) and interactions among multiple biotic and abiotic factors. Future progress will rely on combining global syntheses with coordinated, cross-system experiments that explicitly test how biodiversity scales through space, time and function. Priorities include creating hybrid experimental and observational networks that embed manipulations within long-term monitoring frameworks (for example, plots scale manipulations on reef patches and survey data collected at both local and regional scales); hierarchical sampling designs to quantify α-, β- and γ-diversity contributions to service delivery; and standardized experimentation across ecosystem types to elucidate mechanisms, generalities and context-dependant differences. Predicting BES and BEF relationships In the effect size analysis above, we demonstrate that multiple ecosystem services and functions, including carbon sequestration, biomass turnover rates and food and feed provision, exhibit significant sensitivity to biodiversity loss. Using these data, we predict future shifts in biodiversity relationships with a key ecosystem service. Existing literature is geographically biased towards developed countries (Fig. 1a ), underrepresenting regions critical for global food production, such as South America and Africa 50 . To address this gap, we developed a novel heuristic modelling framework that links global BES and BEF relationships with projected changes in land use and associated anthropogenic pressures. By integrating socioeconomic scenario-based biodiversity projections from the Biodiversity Intactness Index (BII; derived under the Intergovernmental Panel on Climate Change (IPCC) Shared Socioeconomic Pathways (SSPs)) 51 , 52 , and our database, we generate spatially explicit projections of biodiversity-mediated changes in ecosystem service delivery. This framework enables identification of regions where biodiversity loss may most strongly affect ecosystem functioning. We apply our modelling framework to a case study on agricultural pest regulation, a globally critical issue for food security. This focus aligns with the predominance of terrestrial data in our database and the growing attention on agricultural sustainability 53 , 54 . In particular, we examined the ratio of natural enemies to pests, a widely used proxy for top-down biological control across heterogeneous ecological systems 53 , 54 , 55 . We forecast changes in this ratio as an integrated measure of biological pest regulation rather than modelling changes in natural enemies or pest populations separately. Top-down biological control of agricultural pests by natural enemies supports crop production through the regulation of pest populations and is recognized as a key regulating Nature’s Contribution to People in global assessments by IPBES 11 . Accordingly, changes in predator–pest dynamics represent an important mechanism through which biodiversity influences agricultural production under global environmental change. To link our BES model with large-scale projections of biodiversity, we scaled BES data to BII 56 data to estimate future changes in the ratio of natural enemies to pests and associated pest control services. The BII projections capture biodiversity responses to anthropogenic pressures, primarily land-use change, under different socioeconomic scenarios. Accordingly, projected regional differences primarily reflect variation in land-use intensity, biodiversity intactness and socioeconomic vulnerability embedded within SSP scenarios rather than direct climate effects on pest or predator dynamics. We find that regions with rapid population growth are projected to experience greater pressure on agricultural systems and stronger declines in biodiversity-mediated pest regulation. We focus on two SSPs 51 : SSP2, a ‘middle-of-the-road’ trajectory reflecting continuation of current socioeconomic trends, and SSP5, a fossil fuel-driven trajectory characterized by intensive resource exploitation. Our projections suggest that biological pest regulation services relevant to food and feed production are reduced under fossil fuel development compared with the middle-of-the-road development trajectory, though the impacts are heterogeneous (Fig. 4a,b and Supplementary Fig. 2 ) and particularly pronounced in regions experiencing rapid population growth and with low Human Development Index (HDI) values (Fig. 4c,d and Supplementary Table 2 ). HDI is a composite measure of national development based on life expectancy, education and income. Countries with low HDI typically experience higher reliance on local ecosystem services and more rapid population growth compared with high HDI countries. As such, HDI provides a useful proxy for socioeconomic vulnerability and dependence on biodiversity-mediated ecosystem services 57 . a , b , Here, we combine our BES data with two IPCC shared sociopolitical climate change scenarios (SSPs) 45 and their resulting biodiversity intactness values 46 under middle-of-the-road ( a ) (SSP2) and fossil-fuel development ( b ) (SSP5) scenarios (see Supplementary Fig. 2 for all scenarios). While the differences between these two scenarios for biodiversity are complex, our model forecasts that under each scenario, there will be subtly different degrees of protection provided from pests by natural enemies ( a and b ). c , d , Results from two-sided SAR regression models accounting for spatial autocorrelation between regions ( Methods ). A decrease in natural protection from pests would be expected to make agricultural production more difficult, and these changes will disproportionately affect countries experiencing greater population growth and with low HDI values ( c and d ). No adjustment for multiple comparisons was applied. Predicted relative changes in food and feed at the country level between 1970 and 2050 are plotted against the percentage change in population between 1970 and 2050 ( c ) and the human development index, HDI ( d ), for both SSP2 and SSP5. We found that pest protection is reduced under fossil fuel-driven development (SSP5; β = –0.0488 ± 0.0105, z = –4.65, P = 3.34 × 10 −6 ), and we observe a disproportionately greater decline in faster-growing countries ( β = –0.0088 ± 0.0021, z = –4.28, P = 1.85 × 10 −5 ; Supplementary Table 2 ). Pest protection is also reduced in countries with low HDI values ( β = 0.176 ± 0.038, z = 4.59, P = 4.41 × 10 −6 ) which is further exacerbated under fossil fuel development ( β = –0.049 ± 0.011, z = –4.60, P = 4.27 × 10 −6 ; Supplementary Table 3 ). This case study highlights the importance of considering forecasting not only climate or biodiversity but also ecosystem services when considering future decision-making. World map data from Natural Earth ( https://www.naturalearthdata.com/ ). Our model predicts reductions in the natural enemy to pest ratio under SSP5 relative to SSP2 in North America, South America and Asia, whereas Australasia and Africa show comparatively minor differences between scenarios. These results suggest that biodiversity loss may weaken an important ecological pathway underpinning agricultural production, particularly in regions facing high demand for food systems. By contrast, countries with higher HDI values are predicted to experience comparatively greater levels of biological pest regulation potential, reflecting greater capacity to maintain biological pest regulation. An important limitation is that projected BII changes do not account for which species are lost under future scenarios. If species vulnerability is correlated with their functional contributions, future ecosystem service changes may differ from our projections. Future integration of species-level vulnerability and functional traits will therefore be important for assessing how non-random biodiversity change influences BES relationships. Importantly, our framework enables these regional and scenario-based differences to be quantified and compared on a common scale, providing a scientifically grounded basis for assessing biodiversity-driven changes in ecosystem services. We release model outputs and forecasts across all 23 ecosystem services and functions and five SSPs in the supplementary materials to support policy and planning. Our analysis, leveraging the largest global synthesis of its kind, demonstrates that biodiversity underpins a wide range of ecosystem functions and ecosystem services, with biomass turnover, carbon sequestration and food and feed provision showing particularly strong sensitivity to high biodiversity. Exceptional sensitivity of oceanic carbon sequestration underscores the urgency of marine conservation. By combining this synthesis with a heuristic modelling framework, we predict how biodiversity loss may alter key ecosystem functions and services under future socioeconomic and land-use scenarios, identifying regions and services most at risk. Distinguishing between functional redundancy and functionally unique species, we show that conservation priorities must balance maintaining overall diversity with safeguarding irreplaceable foundational taxa. Contextual factors, including study type, ecosystem type and spatial grain, modulate biodiversity effects, highlighting the importance of context-dependent patterns. Integrating global syntheses with targeted cross-system experiments, standardized trait and service metrics and hierarchical sampling will be essential to resolve mechanisms and improve predictions of ecosystem service provision under global change. This study presents a comprehensive global analysis of the relationship between biodiversity and ecosystem services and biodiversity ecosystem functions, aiming to elucidate the impacts of biodiversity loss on critical ecological processes worldwide through the synthesis of a substantial body of existing BES and BEF research. We first investigated functional variation in BES and BEF relationships to assess the prevalence of linear versus saturating forms, the results of which underscore the consistent importance of high biodiversity levels. Subsequently, we quantified the magnitude of biodiversity’s influence on various functions, with particular emphasis on vital services such as carbon sequestration. Our analysis then explored factors influencing BES and BEF trends, including spatial grain, ecosystem type and study type, to understand their effects on BES and BEF relationships across diverse ecosystems and research methodologies. Finally, we predicted BES and BEF relationships under projected future scenarios, using IPCC climate projections and biodiversity forecasts to model potential consequences for an ecosystem service critical to human well-being: food provision 58 . All data analysis and statistical modelling were performed using the R programming language (R-4.4.2 59 ), and the complete database is made publicly available. We express gratitude to the numerous researchers whose dedicated efforts and invaluable contributions were fundamental to this global synthesis; a comprehensive list of all contributors is provided in Supplementary Table 4 . Stratified literature survey We downloaded BES and BEF literature from the Web of Science (WoS) using the following search string: TS = (‘ecosystem service*’ OR ‘ecosystem function*’ OR (‘ecosystem stability’ AND ‘biodiversity’) OR ‘species richness’ OR ‘species diversity’). Given the diversity of metrics and terminology used in this field, relevant studies may not always be captured by strictly nested search strings. All records were therefore screened to ensure that only studies explicitly linking biodiversity to ecosystem services or functions were retained. This search resulted in 312,291 publications. To reduce the number of papers while ensuring a representative spread of data, we used a stratified survey method. After first filtering the literature search results by ecosystem type and ecoregion. Within each stratum, records were ranked using the WoS relevance algorithm (based on keyword matching rather than citation counts), and the top 100 studies were selected for screening, resulting in 3,875 papers. Selecting a fixed number of highly relevant studies per ecosystem and ecoregion category allowed consistent screening and detailed data extraction within practical constraints. In addition, we carried out a specific search for Biodiversity OR species richness OR species diversity and each of the top ten most reported ES, as defined by the IPBES 60 , based on a WoS keyword search (energy, regulation of climate, food and feed, regulation of air quality, NPP, maintenance of options, regulation of freshwater and coastal water quality, biomass turnover rate and evapotranspiration). We added the top 100 most relevant papers for each of these ten ecosystem services categories to our literature database, meaning our stratified survey resulted in 4,875 papers (Supplementary Fig. 3 ). When adding studies to our database, study relevance was not determined on the basis of study outcomes or statistical significance but strictly on the basis of predefined inclusion criteria. Studies were included if they (1) explicitly quantified biodiversity (for example species richness, diversity indices, functional or phylogenetic diversity) and (2) reported empirical relationships with at least one ecosystem function or service metric. Studies were excluded if they were purely theoretical, did not include quantitative biodiversity measures or did not report a direct BEF/BES relationship. Where multiple results were available within a study (for example, across ecosystem types, regions or services), all eligible effect sizes that met the inclusion criteria were extracted (Supplementary Fig. 3 ). As our literature review method was not exhaustive, it was not designed to quantitatively measure biases in data collection across ecosystem type or ES. However, we would not expect the paper-selection processes to select certain systems over others, as data across regions and ecosystem type was explicitly considered. Our literature sorting was intended to be blind with respect to geographic biases, but we acknowledge that, given the geographic biases shown in Fig. 1 , these may have crept into our analysis. We suggest, however, that the distribution of data most likely represents the underlying biases in funding and research into BES and BEF relationships. For example, it is quite clear from Fig. 1 that there is a bias towards Europe, and particularly Germany. We think it is unlikely that there is a reason that Germany should be highlighted by our search terms, instead its prominence likley reflects the presence of a number of key experiments (for example, Jena, BIODEPTH 22 , 61 ). While most of the ecosystem services categories reported are commonly used in the biodiversity literature, some are less well known such as ‘maintenance of options’, which refers to the ‘capacity of ecosystems, habitats, species or genotypes to keep options open to support a good quality of life’, or ‘biological pump efficiencies’, which concerns the transport of carbon from surface water into deeper water and sediments where the residence time is much longer 60 . The abstract and keywords of each paper in our stratified survey were screened for BES and BEF data. If a paper was relevant, we extracted raw data where possible; if raw data were unavailable, we extracted average and standard deviation values for each condition. We used Web Plot Digitiser, in-text tables, supplementary information or online data repositories to extract BES and BEF data. Papers whose data were unavailable were discarded (Supplementary Fig. 3 ). From each of the papers included in our database, we collected data on environmental context (ecosystem type, location, experimental treatment, land use, GPS, spatial grain or the size of the unit of measurement) and any experimental manipulation (that is, temperature and biodiversity) and measurement type (experimental and observational). Land-use categories were defined by IPBES 54 . In total, our database includes data from 545 papers. All analyses were carried out using R (v. 4.4.2). Frequently, a given study would measure multiple functions or services, measure across different experiments within a broader study or within sites within a wider region. To separate and distinguish among these groups so that we could assess those, while also being able to statistically control for differences among studies, we assigned studies with multiple measurements of biodiversity, ecosystem service or function, location or taxon dataset numbers. We use these dataset numbers for analysis. Within each dataset, we filtered out BES and BEF data with fewer than six data points for either ecosystem service or function or biodiversity measurement. After filtering, our database includes 222,829 data points from 1,959 unique datasets extracted from 423 unique papers. The study selection and screening process is summarized in detail in the PRISMA flow diagram (Supplementary Fig. 3 ). Data were standardized to control for different measurement units across studies using the scale function in R, with the standardized data used to calculate Akaike information criterion (AIC) scores and effect size values (see below). Rescaling our biodiversity and ecosystem data allowed us to control for differences in measurement methods and to enable a greater conceptual understanding. First, studies use diverse metrics (for example, species richness, evenness and functional and phylogenetic diversity), as well as differing sampling effort, spatial scale and regional species pools. This heterogeneity impedes direct comparison of raw biodiversity values, introducing both statistical noise and interpretive ambiguity in cross-system syntheses. Thus, rescaling isolates the within-study relationship between biodiversity and ecosystem service and functions, enabling comparability across studies despite differences in biodiversity measurements and sampling design, which was critical to identifying general BES and BEF patterns across ecosystems and experimental contexts. This approach also lets us summarize and unpick how systems respond to differences in biodiversity seen within that system, be it observational or experimental, and so understand how ecosystems might be expected to respond. BES and BEF form To determine the form (shape) of BES and BEF relationships, we applied multiple statistical models to evaluate if relationships were linear or saturating (quadratic or logarithmic). Model selection was based on Akaike weights, identifying the best-fitting relationship for each dataset. We calculated a single coefficient in the case of linear and logarithmic relationships and two coefficients in the case of polynomial relationships. We calculated AIC scores for BES and BEF data fit by linear, quadratic and logarithmic models for each dataset. Using these AIC scores, we calculated Akaike weights for each dataset to determine the relative likelihood that a model is correct compared with the other models considered. We calculated the average Akaike weight across datasets for each ecosystem service and function category for comparison. Saturation types We assessed saturation points of BES and BEF relationships across ecosystem types. Saturation points were defined as the biodiversity value at which the slope of the relationship reached zero for quadratic models with negative, significant curvature or the value at which the slope declined to 10% of its initial value for logarithmic models; linear models were treated as non-saturating. Analyses were done on non-standardized data and restricted to taxonomic, count-based species richness data. Variation in biodiversity metrics and relationship forms To assess how methodological choices vary across ecosystem service and ecosystem function categories, we analysed both the distribution of biodiversity metric classes and the form of fitted BEF and BES relationships. Biodiversity metrics were grouped into four classes (taxonomic, functional, phylogenetic and other). Our literature search was designed to capture broad BEF and BES relationships rather than systematically sample biodiversity metric types; as such, these analyses should be interpreted as descriptive patterns within the compiled literature rather than unbiased estimates of metric usage across the field. First, we tested whether the distribution of biodiversity metric classes differed among ecosystem service and ecosystem function categories using a chi-squared test of independence, based on the contingency table shown in Extended Data Fig. 1 . Because several cells had low expected frequencies, significance was assessed using Monte Carlo simulation (9,999 permutations). Effect size was quantified using Cramér’s V , and standardized residuals were used to identify over- and underrepresented combinations. We then examined whether best-fitting relationship form differed among biodiversity metric classes. For each observation, the best-fitting model (linear, logarithmic or quadratic) was selected on the basis of the highest Akaike weight. A contingency table of biodiversity metric class versus model type was analysed using a chi-squared test, and standardized residuals were used to identify deviations from expected frequencies. To determine differences in biodiversity effects among ecosystem services and functions, we calculated effect size as the response of services and functions to increasing biodiversity for each dataset. To calculate effect size, we used simple correlation coefficients ( r ) between BES relationships, which were normalized using Fisher’s z -algorithm 34 (Zr) as Zr normalizes correlation coefficients for statistical comparison and allows us to compare the average strength of BES and BEF relationships among categories. We use the metafor package (v. 4.2-0) to carry out Cochran’s Q test and construct a funnel plot to test for residual heterogeneity. Ecosystem type, study type and spatial grain We used a reduced linear mixed-effects model to determine the effects of ecosystem service and function category, ecosystem type, study type and spatial grain on BES and BEF relationships. We explicitly included interactions between biodiversity and each of these factors to test whether the strength or direction of BES and BEF relationships depended on ecosystem function or service type, biome, study design or spatial scale. In additiont, three-way interactions between biodiversity, ecosystem service type, and each factor were included to capture more complex, function-specific dependencies. We used the dataset as a random effect to control for differences across studies and this was forced to be in every model during model selection. AIC-based model selection was then used to retain the most informative effects while minimizing overfitting. We did not run model averaging, as no other model was well supported by our data (ΔAICc <4). Our final model was: ecosystem service value ≈ biodiversity value + ecosystem type + ecosystem service category + study type + spatial grain + (1|dataset) + biodiversity value:ecosystem type + biodiversity value:ecosystem service category + biodiversity value:study type + biodiversity value:spatial grain + biodiversity value:ecosystem type:ecosystem service category + biodiversity value:ecosystem service category:study type + biodiversity value:ecosystem service category:spatial grain. Models were run using lme4 and lmerTest packages, and model selection was performed using MuMIn in R 62 , 63 , 64 . Forecast future changes to ecosystem service To model-predicted future changes to biodiversity, we used projected BII values from the PREDICTS project 56 , 65 . BII is the average abundance across a large and functionally diverse set of species relative to their reference populations but does not consider climate change impact 52 . Specifically, we obtained rasterized (spatial resolution 0.25 degrees) predicted BII values (and country-level averaged BII values) for each of the five SSP scenarios described by the IPCC 66 , 67 . These pathways include sustainable development (SSP1), middle of the road (SSP2), regional rivalry (SSP3), inequality (SSP4) and fossil fuel development (SSP5) and were taken for the following years: 1970, 2020, 2030, 2040 and 2050. We chose these pathways as they represent the most realistic alternative socioeconomic development trajectories under future socioeconomic development 66 . There is a fundamental disconnect between how biodiversity is measured and how potential future changes to biodiversity are forecast. The latter typically comprises unitless, bounded indices (such as BII), whereas the former comprises a variety of measurements, including primary production, species abundance and percentage cove 7 , 68 . These metrics vary substantially in terms of their units of measurement, spatial scale, and the range of plausible data values. To address this issue, matching the range of values of BII data (bounded from 0–1.18) and BES data, we used four different scaling procedures, so biodiversity values correspond to BII values of 0–1.18, 0.3–1.18, 0–0.6 and 0.3–0.6, respectively. These ranges were chosen to capture the full spectrum of possible values, exclude extreme low outliers, focus on low-to-moderate biodiversity or highlight the mid-range where most sites cluster. Using multiple overlapping scales ensures that observed trends in BES relationships are robust and not an artefact of the chosen scaling procedure, allowing clear visualization of qualitative patterns. This was achieved by applying a scaling function as follows: where x is an unscaled biodiversity value, u is the upper bound of the scaled data (1, 0.6 or 0.6), and \(l\) is the lower bound of the scaled data (0.3, 0 or 0.3). While this does not create exact quantitative equivalence between metrics, it ensures that trends in BES relationships can be visualized unambiguously and compared across different types of biodiversity data, allowing qualitative patterns to emerge despite the diversity of original measurements. All non-terrestrial biodiversity and ecosystem services data were excluded, as PREDICTS contains BII values for terrestrial ecosystems only. To control for baseline differences in ES, BII data were subset by ecoregions to allow ecosystem services predictions to be made separately for each ecoregion 69 . Within our database, each BES data point was also assigned an ecoregion. Predictions for future ecosystem services values were obtained by inputting scaled BII values into BES forecast models using a posterior sampling method. Thus, projected changes represent changes in the measured ecosystem service response variable used in each BES relationship, rather than independent predictions of the underlying biological components contributing to that service. Our projections assume that changes in BII are independent of the functional importance of the species affected. Consequently, projected changes in ecosystem services reflect aggregate changes in biodiversity rather than species-specific compositional turnover or differential contributions of individual species to ecosystem functioning. As the observed BES relationships are underlain by uncertainty owing to variation in experimental approaches, spatiotemporal scales and sample sizes, we used a Monte Carlo uncertainty approach to propagate this uncertainty forward. Specifically, one hundred predictions of future ecosystem services were drawn from two normal distributions (one centred around the mean BES coefficient and the other centred around the mean intercept, both with standard error equivalent to the mean standard error of BES coefficients and intercepts respectively) for each unique combination of ES, ecoregion, year and SSP. In the case of quadratic relationships, three normal distributions were used for posterior sampling owing to the existence of a quadratic BES coefficient in addition to the linear BES coefficient term. The predictions obtained by this procedure are, at this stage, relative and unitless. This makes inferring the scale of the predicted change in ecosystem services challenging. Consequently, we scaled predicted ecosystem services values to represent absolute values with known units as follows: where p is the predicted ecosystem service at a BII value x , and f min and f max are the raw, absolute minimum and maximum ecosystem service values. As ecosystem service values do not have the same units, f min and f max were obtained from the ecosystem service type that was most prevalent for each given ecosystem service. To visualize ecosystem service forecasts, mean predicted ecosystem service values were generated for each unique combination of ecosystem service, ecoregion, year and SSP, and mean forecasts were plotted using the R package sf. To visualize spatial uncertainty in predicted ecosystem services, Monte Carlo simulations for each combination of ecosystem service type, ecoregion, year and climate scenario were compiled into raster stacks. From these stacks, we calculated 95% credible interval (a Bayesian measure of uncertainty around estimated values) at each grid cell as the 2.5th and 97.5th percentiles of the posterior predictive distribution. These intervals were mapped alongside mean predicted ecosystem service values to produce spatially explicit projections with associated uncertainty. This approach allows the comparison of both expected values and uncertainty across ecoregions and climate scenarios, providing a robust and interpretable representation of projected BEF and BES dynamics. Country-level predictions, population change and HDI We examined the potential for correlation between future population change or HDI against forecasted BES relationships. HDI is a composite indicator based on life expectancy, education and income, with high values reflecting greater socioeconomic development and typically greater capacity to buffer environmental change, whereas low values indicate lower development and higher reliance on local ecosystem services. We used the approach outlined in the section above (‘Forecast future changes to ecosystem service’ section) to forecast country-level relative changes in ‘food and feed’ provisioning in the years 1970 and 2050 under SSP2 and SSP5 pathways. For these forecasts, country-level average BII values were used. The percentage change in population forecasted to occur between 1970 and 2050 in each country was obtained from the United Nations Department of Economic and Social Affairs Population Division 70 . HDI data, which is a summary of three dimensions of human health (a long and healthy life, being knowledgeable and having a decent standard of living) are from 2023, the most recent year available 57 . Note that any territory for which BII data were not accessible was excluded. Projected regional differences in ecosystem service outcomes emerge from the interaction between scenario-driven changes in biodiversity (via BII under SSPs), socioeconomic context (captured by HDI and population trajectories) and uncertainty in the underlying BES relationships, which is propagated through all projections. We assessed the association between projected population growth, HDI and the relative change in food and feed provisioning between 1970 and 2050. We used a spatial autoregressive (SAR) model to account for spatial autocorrelation between regions in both models (percentage population growth, Moran’s I = 0.457, P < 0.0001; HDI, Moran’s I = 0.449, P < 0.0001). We use the R package spatialreg to construct the SAR model 71 . Assumptions and limitations of predictive modelling This predictive framework does not explicitly represent key mediating factors known to influence pest regulation outcomes, including landscape configuration, crop type, agricultural management practices or climate-driven changes in crop productivity. The model therefore represents biodiversity-only effects on a biotic regulatory pathway rather than a full agricultural yield or food system model. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Data availability The datasets generated during and analysed during the current study are available via Zenodo at https://doi.org/10.5281/zenodo.21456540 (ref. 72 ). Source data are provided with this paper. Code availability The code generated to produce models in this study is available via Zenodo at https://doi.org/10.5281/zenodo.21456540 (ref. 72 ). Brose, U. & Hillebrand, H. Biodiversity and ecosystem functioning in dynamic landscapes. Philos. Trans. R. Soc. Lond. B 371 , 20150267 (2016). Article Google Scholar Brose, U. & Hillebrand, H. Biodiversity and ecosystem functioning in dynamic landscapes. Philos. Trans. R. Soc. Lond. B 371 , 20150267 (2016). Article Google Scholar Isbell, F. et al. Linking the influence and dependence of people on biodiversity across scales. Nature 546 , 65–72 (2017). Article CAS PubMed PubMed Central Google Scholar Isbell, F. et al. Linking the influence and dependence of people on biodiversity across scales. Nature 546 , 65–72 (2017). Article CAS PubMed PubMed Central Google Scholar Cardinale, B. et al. Biodiversity loss and its impact on humanity. Nature 486 , 59–67 (2012). Article CAS PubMed Google Scholar Cardinale, B. et al. Biodiversity loss and its impact on humanity. Nature 486 , 59–67 (2012). Article CAS PubMed Google Scholar Díaz, S. et al. Pervasive human-driven decline of life on Earth points to the need for transformative change. Science 366 , eaax3100 (2019). Article PubMed Google Scholar Díaz, S. et al. Pervasive human-driven decline of life on Earth points to the need for transformative change. Science 366 , eaax3100 (2019). Article PubMed Google Scholar Newbold, T. et al. Global effects of land use on local terrestrial biodiversity. Nature 520 , 45–50 (2015). Article CAS PubMed Google Scholar Newbold, T. et al. Global effects of land use on local terrestrial biodiversity. Nature 520 , 45–50 (2015). Article CAS PubMed Google Scholar Pereira, H. M. Global trends and scenarios for terrestrial biodiversity and ecosystem services from 1900 to 2050. Science 384 , 458–465 (2024). Article CAS PubMed Google Scholar Pereira, H. M. Global trends and scenarios for terrestrial biodiversity and ecosystem services from 1900 to 2050. Science 384 , 458–465 (2024). Article CAS PubMed Google Scholar Purvis, A. & Hector, A. Getting the measure of biodiversity. Nature 405 , 212–219 (2000). Article CAS PubMed Google Scholar Purvis, A. & Hector, A. Getting the measure of biodiversity. Nature 405 , 212–219 (2000). Article CAS PubMed Google Scholar Dasgupta, P. The Economics of Biodiversity: The Dasgupta Review: Full Report (HM Treasury, 2021). Dasgupta, P. The Economics of Biodiversity: The Dasgupta Review: Full Report (HM Treasury, 2021). Joly, C. A. The Kunming-Montréal Global Biodiversity Framework. Biota Neotrop. 22 , e2022e001 (2023). Article Google Scholar Joly, C. A. The Kunming-Montréal Global Biodiversity Framework. Biota Neotrop. 22 , e2022e001 (2023). Article Google Scholar Biodiversity strategy for 2030. European Commission https://environment.ec.europa.eu/strategy/biodiversity-strategy-2030_en (2025). Biodiversity strategy for 2030. European Commission https://environment.ec.europa.eu/strategy/biodiversity-strategy-2030_en (2025). IPBES. Global Assessment Report on Biodiversity and Ecosystem Services of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services. Zenodo https://doi.org/10.5281/zenodo.6417333 (2019). IPBES. Global Assessment Report on Biodiversity and Ecosystem Services of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services. Zenodo https://doi.org/10.5281/zenodo.6417333 (2019). Garland, G. et al. A closer look at the functions behind ecosystem multifunctionality: a review. J. Ecol. 109 , 600–613 (2021). Article Google Scholar Garland, G. et al. A closer look at the functions behind ecosystem multifunctionality: a review. J. Ecol. 109 , 600–613 (2021). Article Google Scholar Costanza, R., Fisher, B., Mulder, K., Liu, S. & Christopher, T. Biodiversity and ecosystem services: a multi-scale empirical study of the relationship between species richness and net primary production. Ecol. Econ. 61 , 478–491 (2007). Article Google Scholar Costanza, R., Fisher, B., Mulder, K., Liu, S. & Christopher, T. Biodiversity and ecosystem services: a multi-scale empirical study of the relationship between species richness and net primary production. Ecol. Econ. 61 , 478–491 (2007). Article Google Scholar Sun, G., Hallema, D. & Asbjornsen, H. Ecohydrological processes and ecosystem services in the Anthropocene: a review. Ecol. Process. 6 , 35 (2017). Article Google Scholar Sun, G., Hallema, D. & Asbjornsen, H. Ecohydrological processes and ecosystem services in the Anthropocene: a review. Ecol. Process. 6 , 35 (2017). Article Google Scholar Amelung, W. et al. Towards a global-scale soil climate mitigation strategy. Nat. Commun. 11 , 5427 (2020). Article CAS PubMed PubMed Central Google Scholar Amelung, W. et al. Towards a global-scale soil climate mitigation strategy. Nat. Commun. 11 , 5427 (2020). Article CAS PubMed PubMed Central Google Scholar Schwartz, M. W. et al. Linking biodiversity to ecosystem function: implications for conservation ecology. Oecologia 122 , 297–305 (2000). Article CAS PubMed Google Scholar Schwartz, M. W. et al. Linking biodiversity to ecosystem function: implications for conservation ecology. Oecologia 122 , 297–305 (2000). Article CAS PubMed Google Scholar Wagg, C. et al. Diversity and asynchrony in soil microbial communities stabilizes ecosystem functioning. Elife 10 , e62813 (2021). Article CAS PubMed PubMed Central Google Scholar Wagg, C. et al. Diversity and asynchrony in soil microbial communities stabilizes ecosystem functioning. Elife 10 , e62813 (2021). Article CAS PubMed PubMed Central Google Scholar Tilman, D., Reich, P. & Knops, J. Biodiversity and ecosystem stability in a decade-long grassland experiment. Nature 441 , 629–632 (2006). Article CAS PubMed Google Scholar Tilman, D., Reich, P. & Knops, J. Biodiversity and ecosystem stability in a decade-long grassland experiment. Nature 441 , 629–632 (2006). Article CAS PubMed Google Scholar Tucker, C. M. et al. Assessing the utility of conserving evolutionary history. Biol. Rev. 94 , 1740–1760 (2019). Article PubMed PubMed Central Google Scholar Tucker, C. M. et al. Assessing the utility of conserving evolutionary history. Biol. Rev. 94 , 1740–1760 (2019). Article PubMed PubMed Central Google Scholar Cadotte, M. W. Functional traits explain ecosystem function through opposing mechanisms. Ecol. Lett. 20 , 989–996 (2017). Article PubMed Google Scholar Cadotte, M. W. Functional traits explain ecosystem function through opposing mechanisms. Ecol. Lett. 20 , 989–996 (2017). Article PubMed Google Scholar Loreau, M. & Hector, A. Partitioning selection and complementarity in biodiversity experiments. Nature 412 , 72–76 (2001). Article CAS PubMed Google Scholar Loreau, M. & Hector, A. Partitioning selection and complementarity in biodiversity experiments. Nature 412 , 72–76 (2001). Article CAS PubMed Google Scholar Hossain, M. L., Li, J., Hoffmann, S. & Beierkuhnlein, C. Biodiversity showed positive effects on resistance but mixed effects on resilience to climatic extremes in a long-term grassland experiment. Sci. Total Environ. 827 , 154322 (2022). Article CAS PubMed Google Scholar Hossain, M. L., Li, J., Hoffmann, S. & Beierkuhnlein, C. Biodiversity showed positive effects on resistance but mixed effects on resilience to climatic extremes in a long-term grassland experiment. Sci. Total Environ. 827 , 154322 (2022). Article CAS PubMed Google Scholar Schmid, B. et al. Consequences of species loss for ecosystem functioning: meta-analyses of data from biodiversity experiments. in Biodiversity, Ecosystem Functioning, and Human Wellbeing: An Ecological and Economic Perspective (eds Naeem, S. et al.) 14–29 https://doi.org/10.1093/acprof:oso/9780199547951.003.0002 (Oxford Univ. Press, 2009). Schmid, B. et al. Consequences of species loss for ecosystem functioning: meta-analyses of data from biodiversity experiments. in Biodiversity, Ecosystem Functioning, and Human Wellbeing: An Ecological and Economic Perspective (eds Naeem, S. et al.) 14–29 https://doi.org/10.1093/acprof:oso/9780199547951.003.0002 (Oxford Univ. Press, 2009). Ren, Y., Lü, Y., Fu, B. & Zhang, K. Biodiversity and ecosystem functional enhancement by forest restoration: a meta-analysis in China. Land Degrad. Dev. 28 , 2062–2073 (2017). Article Google Scholar Ren, Y., Lü, Y., Fu, B. & Zhang, K. Biodiversity and ecosystem functional enhancement by forest restoration: a meta-analysis in China. Land Degrad. Dev. 28 , 2062–2073 (2017). Article Google Scholar van der Plas, F. Biodiversity and ecosystem functioning in naturally assembled communities. Biol. Rev. 94 , 1220–1245 (2019). Article PubMed Google Scholar van der Plas, F. Biodiversity and ecosystem functioning in naturally assembled communities. Biol. Rev. 94 , 1220–1245 (2019). Article PubMed Google Scholar Hooper, D. et al. Effects of biodiversity on ecosystem functioning: a consensus of current knowledge. Ecol. Monogr. 75 , 3–35 (2005). Article Google Scholar Hooper, D. et al. Effects of biodiversity on ecosystem functioning: a consensus of current knowledge. Ecol. Monogr. 75 , 3–35 (2005). Article Google Scholar Cardinale, B., Bennett, D., Nelson, C. & Gross, K. Does productivity drive diversity or vice versa? A test of the multivariate productivity-diversity hypothesis in streams. Ecology 90 , 1227–1241 (2009). Article PubMed Google Scholar Cardinale, B., Bennett, D., Nelson, C. & Gross, K. Does productivity drive diversity or vice versa? A test of the multivariate productivity-diversity hypothesis in streams. Ecology 90 , 1227–1241 (2009). Article PubMed Google Scholar Tilman, D., Lehman, C. L. & Thomson, K. T. Plant diversity and ecosystem productivity: theoretical considerations. Proc. Natl Acad. Sci. USA 94 , 1857–1861 (1997). Article CAS PubMed PubMed Central Google Scholar Tilman, D., Lehman, C. L. & Thomson, K. T. Plant diversity and ecosystem productivity: theoretical considerations. Proc. Natl Acad. Sci. USA 94 , 1857–1861 (1997). Article CAS PubMed PubMed Central Google Scholar Gessner, M. O. et al. Diversity meets decomposition. Trends Ecol. Evol. 25 , 372–380 (2010). Article PubMed Google Scholar Gessner, M. O. et al. Diversity meets decomposition. Trends Ecol. Evol. 25 , 372–380 (2010). Article PubMed Google Scholar Worm, B. et al. Impacts of biodiversity loss on ocean ecosystem services. Science 314 , 787–790 (2006). Article CAS PubMed Google Scholar Worm, B. et al. Impacts of biodiversity loss on ocean ecosystem services. Science 314 , 787–790 (2006). Article CAS PubMed Google Scholar Stachowicz, J. J., Bruno, J. F. & Duffy, J. E. Understanding the effects of marine biodiversity on communities and ecosystems. Annu. Rev. Ecol. Evol. Syst. 38 , 739–766 (2007). Article Google Scholar Stachowicz, J. J., Bruno, J. F. & Duffy, J. E. Understanding the effects of marine biodiversity on communities and ecosystems. Annu. Rev. Ecol. Evol. Syst. 38 , 739–766 (2007). Article Google Scholar Daam, M. A., Teixeira, H., Lillebø, A. I. & Nogueira, A. J. A. Establishing causal links between aquatic biodiversity and ecosystem functioning: Status and research needs. Sci. Total Environ. 656 , 1145–1156 (2019). Article PubMed Google Scholar Daam, M. A., Teixeira, H., Lillebø, A. I. & Nogueira, A. J. A. Establishing causal links between aquatic biodiversity and ecosystem functioning: Status and research needs. Sci. Total Environ. 656 , 1145–1156 (2019). Article PubMed Google Scholar Gonzalez, A. et al. Scaling-up biodiversity-ecosystem functioning research. Ecol. Lett. 23 , 757–776 (2020). Article PubMed PubMed Central Google Scholar Gonzalez, A. et al. Scaling-up biodiversity-ecosystem functioning research. Ecol. Lett. 23 , 757–776 (2020). Article PubMed PubMed Central Google Scholar Rosenberg, M. S., Rothstein, H. R. & Gurevitch, J. Effect sizes. in Handbook of Meta-analysis in Ecology and Evolution (eds Gurevitch, J., Koricheva, J. & Mengersen, K.) 61–71 (Princeton Univ. Press, 2013). Rosenberg, M. S., Rothstein, H. R. & Gurevitch, J. Effect sizes. in Handbook of Meta-analysis in Ecology and Evolution (eds Gurevitch, J., Koricheva, J. & Mengersen, K.) 61–71 (Princeton Univ. Press, 2013). Friedlingstein, P. et al. Global carbon budget 2024. Earth Syst. Sci. Data 17 , 965–1039 (2025). Article Google Scholar Friedlingstein, P. et al. Global carbon budget 2024. Earth Syst. Sci. Data 17 , 965–1039 (2025). Article Google Scholar Kemppinen, K. M. S. et al. Global reforestation and biodiversity conservation. Conserv. Biol. 34 , 1221–1228 (2020). Article PubMed Google Scholar Kemppinen, K. M. S. et al. Global reforestation and biodiversity conservation. Conserv. Biol. 34 , 1221–1228 (2020). Article PubMed Google Scholar Hulvey, K. et al. Benefits of tree mixes in carbon plantings. Nat. Clim. Change 3 , 869–874 (2013). Article CAS Google Scholar Hulvey, K. et al. Benefits of tree mixes in carbon plantings. Nat. Clim. Change 3 , 869–874 (2013). Article CAS Google Scholar Duffy, J. E. Why biodiversity is important to the functioning of real-world ecosystems. Front. Ecol. Environ. 7 , 437–444 (2009). Article Google Scholar Duffy, J. E. Why biodiversity is important to the functioning of real-world ecosystems. Front. Ecol. Environ. 7 , 437–444 (2009). Article Google Scholar Klaus, V. H. et al. Do biodiversity-ecosystem functioning experiments inform stakeholders how to simultaneously conserve biodiversity and increase ecosystem service provisioning in grasslands? Biol. Conserv. 245 , 108552 (2020). Article Google Scholar Klaus, V. H. et al. Do biodiversity-ecosystem functioning experiments inform stakeholders how to simultaneously conserve biodiversity and increase ecosystem service provisioning in grasslands? Biol. Conserv. 245 , 108552 (2020). Article Google Scholar Arkema, K. K. et al. Coastal habitats shield people and property from sea-level rise and storms. Nat. Clim. Change 3 , 913–918 (2013). Article Google Scholar

Arkema, K. K. et al. Coastal habitats shield people and property from sea-level rise and storms. Nat. Clim. Change 3 , 913–918 (2013). Article Google Scholar Craven, D. et al. Multiple facets of biodiversity drive the diversity-stability relationship. Nat. Ecol. Evol. 2 , 1579–1587 (2018). Article PubMed Google Scholar Craven, D. et al. Multiple facets of biodiversity drive the diversity-stability relationship. Nat. Ecol. Evol. 2 , 1579–1587 (2018). Article PubMed Google Scholar Loreau, M. et al. Biodiversity and ecosystem functioning: current knowledge and future challenges. Science 294 , 804–808 (2001). Article CAS PubMed Google Scholar Loreau, M. et al. Biodiversity and ecosystem functioning: current knowledge and future challenges. Science 294 , 804–808 (2001). Article CAS PubMed Google Scholar Isbell, F. et al. High plant diversity is needed to maintain ecosystem services. Nature 477 , 199–202 (2011). Article CAS PubMed Google Scholar Isbell, F. et al. High plant diversity is needed to maintain ecosystem services. Nature 477 , 199–202 (2011). Article CAS PubMed Google Scholar Mace, G. M. et al. Approaches to defining a planetary boundary for biodiversity. Glob. Environ. Change 28 , 289–297 (2014). Article Google Scholar Mace, G. M. et al. Approaches to defining a planetary boundary for biodiversity. Glob. Environ. Change 28 , 289–297 (2014). Article Google Scholar Chisholm, R. A. et al. Scale-dependent relationships between tree species richness and ecosystem function in forests. J. Ecol. 101 , 1214–1224 (2013). Article Google Scholar Chisholm, R. A. et al. Scale-dependent relationships between tree species richness and ecosystem function in forests. J. Ecol. 101 , 1214–1224 (2013). Article Google Scholar Kröel-Dulay, G. et al. Field experiments underestimate aboveground biomass response to drought. Nat. Ecol. Evol. 6 , 540–545 (2022). Article PubMed PubMed Central Google Scholar Kröel-Dulay, G. et al. Field experiments underestimate aboveground biomass response to drought. Nat. Ecol. Evol. 6 , 540–545 (2022). Article PubMed PubMed Central Google Scholar Jochum, M. et al. The results of biodiversity–ecosystem functioning experiments are realistic. Nat. Ecol. Evol. 4 , 1485–1494 (2020). Article PubMed Google Scholar Jochum, M. et al. The results of biodiversity–ecosystem functioning experiments are realistic. Nat. Ecol. Evol. 4 , 1485–1494 (2020). Article PubMed Google Scholar Hagan, J., Vanschoenwinkel, B. & Gamfeldt, L. We should not necessarily expect positive relationships between biodiversity and ecosystem functioning in observational field data. Ecol. Lett. 24 , 2537–2548 (2021). Article PubMed Google Scholar Hagan, J., Vanschoenwinkel, B. & Gamfeldt, L. We should not necessarily expect positive relationships between biodiversity and ecosystem functioning in observational field data. Ecol. Lett. 24 , 2537–2548 (2021). Article PubMed Google Scholar Eisenhauer, N. et al. Biodiversity–ecosystem function experiments reveal the mechanisms underlying the consequences of biodiversity change in real world ecosystems. J. Veg. Sci. 27 , 1061–1070 (2016). Article Google Scholar Eisenhauer, N. et al. Biodiversity–ecosystem function experiments reveal the mechanisms underlying the consequences of biodiversity change in real world ecosystems. J. Veg. Sci. 27 , 1061–1070 (2016). Article Google Scholar Sá, J. C. et al. Low-carbon agriculture in South America to mitigate global climate change and advance food security. Environ. Int. 98 , 102–112 (2017). Article PubMed Google Scholar Sá, J. C. et al. Low-carbon agriculture in South America to mitigate global climate change and advance food security. Environ. Int. 98 , 102–112 (2017). Article PubMed Google Scholar IPCC. Summary for policymakers. in Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change (eds Masson-Delmotte, V. et al.) 3−32 https://doi.org/10.1017/9781009157896.001 (Cambridge Univ. Press, 2021). IPCC. Summary for policymakers. in Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change (eds Masson-Delmotte, V. et al.) 3−32 https://doi.org/10.1017/9781009157896.001 (Cambridge Univ. Press, 2021). Scholes, R. J. & Biggs, R. A biodiversity intactness index. Nature 434 , 45–49 (2005). Article CAS PubMed Google Scholar Scholes, R. J. & Biggs, R. A biodiversity intactness index. Nature 434 , 45–49 (2005). Article CAS PubMed Google Scholar Singh, B. K. et al. Climate change impacts on plant pathogens, food security and paths forward. Nat. Rev. Microbiol. 21 , 640–656 (2023). Article CAS PubMed PubMed Central Google Scholar Singh, B. K. et al. Climate change impacts on plant pathogens, food security and paths forward. Nat. Rev. Microbiol. 21 , 640–656 (2023). Article CAS PubMed PubMed Central Google Scholar Muluneh, M. G. Impact of climate change on biodiversity and food security: a global perspective—a review article. Agric. Food Secur. 10 , 36 (2021). Article Google Scholar Muluneh, M. G. Impact of climate change on biodiversity and food security: a global perspective—a review article. Agric. Food Secur. 10 , 36 (2021). Article Google Scholar Martin, E. A., Reineking, B., Seo, B. & Steffan-Dewenter, I. Natural enemy interactions constrain pest control in complex agricultural landscapes. Proc. Natl Acad. Sci. USA 110 , 5534–5539 (2013). Article CAS PubMed PubMed Central Google Scholar Martin, E. A., Reineking, B., Seo, B. & Steffan-Dewenter, I. Natural enemy interactions constrain pest control in complex agricultural landscapes. Proc. Natl Acad. Sci. USA 110 , 5534–5539 (2013). Article CAS PubMed PubMed Central Google Scholar Hudson, L. N. et al. The database of the PREDICTS (Projecting Responses of Ecological Diversity In Changing Terrestrial Systems) project. Ecol. Evol. 7 , 145–188 (2017). Article PubMed Google Scholar Hudson, L. N. et al. The database of the PREDICTS (Projecting Responses of Ecological Diversity In Changing Terrestrial Systems) project. Ecol. Evol. 7 , 145–188 (2017). Article PubMed Google Scholar UNDP. Human Development Report 2025. UNDP https://report.hdr.undp.org (2025). UNDP. Human Development Report 2025. UNDP https://report.hdr.undp.org (2025). Godfray, H. C. J. et al. Food security: the challenge of feeding 9 billion people. Science 327 , 812–818 (2010). Article CAS PubMed Google Scholar Godfray, H. C. J. et al. Food security: the challenge of feeding 9 billion people. Science 327 , 812–818 (2010). Article CAS PubMed Google Scholar R Core Team R: A Language and Environment for Statistical Computing (R Core Team, 2021). R Core Team R: A Language and Environment for Statistical Computing (R Core Team, 2021). Díaz, S. et al. Assessing nature’s contributions to people. Science 359 , 270–272 (2018). Article PubMed Google Scholar Díaz, S. et al. Assessing nature’s contributions to people. Science 359 , 270–272 (2018). Article PubMed Google Scholar Roscher, C. et al. The role of biodiversity for element cycling and trophic interactions: an experimental approach in a grassland community. Basic Appl. Ecol. 5 , 107–121 (2004). Article Google Scholar Roscher, C. et al. The role of biodiversity for element cycling and trophic interactions: an experimental approach in a grassland community. Basic Appl. Ecol. 5 , 107–121 (2004). Article Google Scholar Bartoń, K. MuMIn: multi-model inference. R package version 1.48.19. CRAN https://CRAN.R-project.org/package=MuMIn (2025). Bartoń, K. MuMIn: multi-model inference. R package version 1.48.19. CRAN https://CRAN.R-project.org/package=MuMIn (2025). Bates, D., Mächler, M., Bolker, B. & Walker, S. Fitting linear mixed-effects models using lme4. J. Stat. Softw. 67 , 1–48 (2015). Article Google Scholar Bates, D., Mächler, M., Bolker, B. & Walker, S. Fitting linear mixed-effects models using lme4. J. Stat. Softw. 67 , 1–48 (2015). Article Google Scholar Kuznetsova, A., Brockhoff, P. B. & Christensen, R. H. B. lmerTest package: tests in linear mixed effects models. J. Stat. Softw. 82 , 1–26 (2017). Article Google Scholar Kuznetsova, A., Brockhoff, P. B. & Christensen, R. H. B. lmerTest package: tests in linear mixed effects models. J. Stat. Softw. 82 , 1–26 (2017). Article Google Scholar Purvis, A. et al. ModelLing and Projecting the Response of Local Terrestrial Biodiversity Worldwide to Land Use and Related Pressures: the PREDICTS project. In Advances in Ecological Research Vol. 58 (eds Bohan, D. A., Dumbrell, A. J., Woodward, G. & Jackson, M.) Chap. 6, 201–241 (Academic Press, 2018). Purvis, A. et al. ModelLing and Projecting the Response of Local Terrestrial Biodiversity Worldwide to Land Use and Related Pressures: the PREDICTS project. In Advances in Ecological Research Vol. 58 (eds Bohan, D. A., Dumbrell, A. J., Woodward, G. & Jackson, M.) Chap. 6, 201–241 (Academic Press, 2018). Strandsbjerg Tristan Pedersen, J. et al. An assessment of the performance of scenarios against historical global emissions for IPCC reports. Global Environ. Change 66 , 102199 (2021). Article Google Scholar Strandsbjerg Tristan Pedersen, J. et al. An assessment of the performance of scenarios against historical global emissions for IPCC reports. Global Environ. Change 66 , 102199 (2021). Article Google Scholar Phillips, H. The Biodiversity Intactness Index—country, region and global-level summaries for the year 1970 to 2050 under various scenarios. Natural History Museum https://doi.org/10.5519/he1eqmg1 (2021). Phillips, H. The Biodiversity Intactness Index—country, region and global-level summaries for the year 1970 to 2050 under various scenarios. Natural History Museum https://doi.org/10.5519/he1eqmg1 (2021). Pavoine, S. & Bonsall, M. B. Measuring biodiversity to explain community assembly: a unified approach. Biol. Rev. Camb. Philos. Soc. 86 , 792–812 (2011). Article CAS PubMed Google Scholar Pavoine, S. & Bonsall, M. B. Measuring biodiversity to explain community assembly: a unified approach. Biol. Rev. Camb. Philos. Soc. 86 , 792–812 (2011). Article CAS PubMed Google Scholar Ferdaña, Z., Bruce, C. & Wyckoff, T. Geospatial Conservation at The Nature Conservancy: 2022 Annual Report & Map Book https://geospatial.tnc.org/pages/annual-report (2023). Ferdaña, Z., Bruce, C. & Wyckoff, T. Geospatial Conservation at The Nature Conservancy: 2022 Annual Report & Map Book https://geospatial.tnc.org/pages/annual-report (2023). World Population Prospects 2024: Summary of Results (United Nations, Department of Economic and Social Affairs, Population Division, 2025). World Population Prospects 2024: Summary of Results (United Nations, Department of Economic and Social Affairs, Population Division, 2025). Pebesma, E. & Bivand, R. Spatial Data Science: With Applications in R 1st edn https://doi.org/10.1201/9780429459016 (CRC, 2023). Pebesma, E. & Bivand, R. Spatial Data Science: With Applications in R 1st edn https://doi.org/10.1201/9780429459016 (CRC, 2023). Moffett, E. R. et al. Code and data for: ‘Biodiversity safeguards ecosystem services and functions worldwide’. Zenodo https://doi.org/10.5281/zenodo.21456541 (2026). Moffett, E. R. et al. Code and data for: ‘Biodiversity safeguards ecosystem services and functions worldwide’. Zenodo https://doi.org/10.5281/zenodo.21456541 (2026). Download references Acknowledgements We express gratitude to the numerous researchers whose dedicated efforts and invaluable contributions were fundamental to this synthesis. We release details of all these studies in Supplementary Table 4 and apologize that we cannot cite them in the main text owing to citation constraints. We thank T. Newbold for comments on an early version of this manuscript. We thank R. Preston-Allen for assisting with data collection. We acknowledge the vital financial support provided by The Hitachi Global Foundation, which made this research possible. WDP and the Pearse lab are supported by Singapore Green Finance Centre (Nature-Related Financial Risks in Southeast Asia), the Alan Turing Institute (FA.04), and grant nos. UKRI BB/Y008766/1, NE/X00547X/1 and NE/X013022/1. Author information Authors and Affiliations The Georgina Mace Centre for the Living Planet, Department of Life Sciences, Imperial College London, Berkshire, UK Emma R. Moffett, Joel H. Gayford, Lexi Chen, Olivia F. Morris, Yuming Shi, Lucy Somekh, Nicole Stasik, Andy Purvis, Guy Woodward & William D. Pearse The Georgina Mace Centre for the Living Planet, Department of Life Sciences, Imperial College London, Berkshire, UK Emma R. Moffett, Joel H. Gayford, Lexi Chen, Olivia F. Morris, Yuming Shi, Lucy Somekh, Nicole Stasik, Andy Purvis, Guy Woodward & William D. Pearse , Department of Geography, King’s College London, London, UK Emma R. Moffett , Department of Geography, King’s College London, London, UK Emma R. Moffett Department of Marine Biology and Aquaculture, James Cook University, Townsville, Queensland, Australia Joel H. Gayford Department of Marine Biology and Aquaculture, James Cook University, Townsville, Queensland, Australia Joel H. Gayford Biodiversity Futures Lab, Natural History Museum, London, UK Andy Purvis Biodiversity Futures Lab, Natural History Museum, London, UK Alan Turing Institute, British Library, London, UK William D. Pearse Alan Turing Institute, British Library, London, UK William D. Pearse Emma R. Moffett View author publications Search author on: PubMed Google Scholar Search author on: PubMed Google Scholar Joel H. Gayford View author publications Search author on: PubMed Google Scholar Search author on: PubMed Google Scholar Lexi Chen View author publications Search author on: PubMed Google Scholar Search author on: PubMed Google Scholar Olivia F. Morris View author publications Search author on: PubMed Google Scholar Search author on: PubMed Google Scholar Yuming Shi View author publications Search author on: PubMed Google Scholar Search author on: PubMed Google Scholar Lucy Somekh View author publications Search author on: PubMed Google Scholar Search author on: PubMed Google Scholar Nicole Stasik View author publications Search author on: PubMed Google Scholar Search author on: PubMed Google Scholar Andy Purvis View author publications Search author on: PubMed Google Scholar Search author on: PubMed Google Scholar Guy Woodward View author publications Search author on: PubMed Google Scholar Search author on: PubMed Google Scholar William D. Pearse View author publications Search author on: PubMed Google Scholar Search author on: PubMed Google Scholar E.R.M, W.D.P. and G.W. designed the project and this study; E.R.M. created the data-collection protocols and database; E.R.M., L.C., Y.S., L.S. and N.S. contributed to data collection; E.R.M. and J.H.G. performed the analyses; E.R.M. wrote the initial manuscript; and J.H.G., L.C., Y.S., L.S., O.F.M., N.S., A.P., G.W. and W.D.P. contributed to editing the manuscript. Corresponding author Correspondence to Emma R. Moffett . Ethics declarations Competing interests The authors declare no competing interests. Peer review information Nature Ecology & Evolution thanks Matthew Mitchell, Ariana Sutton-Grier and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available. Additional information Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Extended Data Fig. 1 Patterns of biodiversity metrics and reported relationships across ecosystem services and functions. Heatmaps show the distribution of observations across ( a ) biodiversity metric classes versus ecosystem services and functions, and ( b ) biodiversity metric classes versus reported relationship types. Colour intensity indicates the proportion of observations within each category, with yellow and orange representing higher proportions. The distribution of biodiversity metrics differed significantly among ecosystem service categories (χ 2 = 78.385, Monte Carlo p < 0.001; Cramér’s V = 0.34), with taxonomic diversity metrics common in our database. Biodiversity-ecosystem service and biodiversity ecosystem-function forms also varied among biodiversity metrics (χ 2 = 23.17, df = 6, p < 0.001), with functional and phylogenetic diversity metrics showing an over-representation of saturating relationships, whereas taxonomic relationships were equally represented across model types. Response variables marked with an asterisk (*) are classified as ecosystem functions (EF), all other response variables are ecosystem services (ES). Extended Data Fig. 2 Biodiversity and ecosystem-service and biodiversity ecosystem function saturation points across ecosystem types. Biodiversity data are untransformed species richness (count data) only. Saturation points occur at lower median species richness in freshwater (3.6 ± 1.99) and terrestrial (14.4 ± 9) systems than in marine ecosystems (22.8 ± 16.5), where higher median values are observed. Uncertainty is expressed as median absolute deviation (MAD). Data are from datasets best predicted by quadratic and logarithmic models, as indicated by Akaike Weight values. Saturation for quadratic models was calculated when the slope reached zero, and for logarithmic models, when the slope was 10% of the initial value. The horizontal dashed line shows the median value for each ecosystem type. Large saturation values may be due to the large spatial scale the data were collected at or due to the diverse fauna at lower levels of the food web ( for example , meiofauna in marine ecosystems). Extended Data Fig. 3 Biodiversity and ecosystem service effect sizes vary across ecosystem types. Data are shown for our database’s five most common ecosystem service categories (panels a – e ). Effect size scores (measured with Fisher’s Z-score; Zr) are averaged (±1 SE) across datasets within each ecosystem service category and ecosystem type, where a high effect size value indicates a stronger effect of biodiversity on ecosystem service compared to a lower value. Dataset n values are shown beside the average effect size value. Overall, our database has high dataset n values for terrestrial studies and low n values for estuarine studies, and this was consistent across all ecosystem service categories. Datasets from marine and freshwater ecosystems were not as common as those from terrestrial studies, and for some ecosystem services (regulation of detrimental organisms and biological processes, maintenance of options), data from these ecosystem types were scarce or missing. Extended Data Fig. 4 Comparison of biodiversity responses across ecosystem service categories, as effect size (Fisher’s Z-score; Zr; Panel a) and the number of studies used to calculate effect size (Panel b ). High effect size values indicate a stronger effect of biodiversity on ecosystem service compared to values. Data are split by study type. The effect sizes which are different (positive vs negative) between study types, are highlighted Response variables marked with an asterisk (*) are classified as ecosystem functions (EF), all other response variables are ecosystem services (ES) 11 . Supplementary information Supplementary Information (download PDF ) Supplementary Figs. 1–3 and Tables 1–5. Reporting Summary (download PDF ) Peer Review File (download PDF ) Supplementary Fig. 1 (download CSV ) Supplementary Fig. 1. Comparison of linear, log and quadratic model fits describing BES relationships across ecosystem function categories. Count ( n ) values for each category are shown in Fig. 2b . Points show mean r 2 values, and error bars indicate ±1 s.e. across studies. Response variables that are classified by the IPBES as ecosystem functions are marked with an asterisk (*) 11 , while ecosystem services are not. Source Data Fig. 1 (download CSV ) GPS, spatial grain experimental and observation data from Fig. 1, unprocessed. Source Data Fig. 2 (download CSV ) AIC weight data, one row per dataset/label in our database. Source Data Fig. 3 (download CSV ) Effect size data including standard error and confidence intervals, one row per dataset/label in our database. Source Data Fig. 4 (download CSV ) Country level predictions of food and feed alongside HDI and population data. Source Data Extended Data Fig.1 (download XLSX ) Data on the distribution of observations across biodiversity metric classes versus ecosystem services and functions and biodiversity metric classes versus reported relationship type. Source Data Extended Data Fig. 2 (download CSV ) Saturation point data for marine, freshwater and terrestrial ecosystems. Source Data Extended Data Fig. 3 (download CSV ) Effect size data split by ecosystem type. Source Data Extended Data Fig. 4 (download CSV ) Effect size data averages across ecosystem function/service categories and study type. Rights and permissions Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . Reprints and permissions About this article Cite this article Moffett, E.R., Gayford, J.H., Chen, L. et al. Biodiversity safeguards ecosystem services and functions worldwide. Nat Ecol Evol (2026). https://doi.org/10.1038/s41559-026-03200-4 Download citation Received : 16 January 2026 Received : 16 January 2026 Accepted : 28 August 2026 Accepted : 28 August 2026 Published : 06 October 2026 Published : 06 October 2026 Version of record : 06 October 2026 Version of record : 06 October 2026 DOI : https://doi.org/10.1038/s41559-026-03200-4 DOI : https://doi.org/10.1038/s41559-026-03200-4 Share this article Anyone you share the following link with will be able to read this content: Sorry, a shareable link is not currently available for this article. Provided by the Springer Nature SharedIt content-sharing initiative Explore articles by subject Guide to authors Editorial policies Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.