Quantifying farm sustainability through the lens of ecological theory.

The achievements of the Green Revolution in meeting the nutritional needs of a growing global population have been won at the expense of unintended consequences for the environment. Some of these negative impacts are now threatening the sustainability of food production through the loss of pollinators and natural enemies of crop pests, the evolution of pesticide resistance, declining soil health and vulnerability to climate change. In the search for farming systems that are sustainable both agronomically and environmentally, alternative approaches have been proposed variously called 'agroecological', 'conservation agriculture', 'regenerative' and 'sustainable intensification'. While the widespread recognition of the need for more sustainable farming is to be welcomed, this has created etymological confusion that has the potential to become a barrier to transformation. There is a need, therefore, for objective criteria to evaluate alternative farming systems and to quantify farm sustainability against multiple outcomes. To help meet this challenge, we reviewed the ecological theories that explain variance in regulating and supporting ecosystem services delivered by biological communities in farmland to identify guiding principles for management change. For each theory, we identified associated system metrics that could be used as proxies for agroecosystem function. We identified five principles derived from ecological theory: (i) provide key habitats for ecosystem service providers; (ii) increase crop and non-crop habitat diversity; (iii) increase edge density: (iv) increase nutrient-use efficiency; and (v) avoid extremes of disturbance. By making published knowledge the foundation of the choice of associated metrics, our aim was to establish a broad consensus for their use in sustainability assessment frameworks. Further analysis of their association with farm-scale data on biological communities and/or ecosystem service delivery would provide additional validation for their selection and support for the underpinning theories.

Ecological restoration and rewilding: two approaches with complementary goals?

As we enter the UN Decade on Ecosystem Restoration (2021-2030) and address the urgent need to protect and restore ecosystems and their ecological functions at large scales, rewilding has been brought into the limelight. Interest in this discipline is thus increasing, with a large number of conceptual scientific papers published in recent years. Increasing enthusiasm has led to discussions and debates in the scientific community about the differences between ecological restoration and rewilding. The main goal of this review is to compare and clarify the position of each field. Our results show that despite some differences (e.g. top-down versus bottom-up and functional versus taxonomic approaches) and notably with distinct goals - recovery of a defined historically determined target ecosystem versus recovery of natural processes with often no target endpoint - ecological restoration and rewilding have a common scope: the recovery of ecosystems following anthropogenic degradation. The goals of ecological restoration and rewilding have expanded with the progress of each field. However, it is unclear whether there is a paradigm shift with ecological restoration moving towards rewilding or vice versa. We underline the complementarity in time and in space of ecological restoration and rewilding. To conclude, we argue that reconciliation of these two fields of nature conservation to ensure complementarity could create a synergy to achieve their common scope.

Correlated biodiversity change between plant and insect assemblages resurveyed after 80 years across a dynamic habitat mosaic.

Historical data on co-occurring taxa are extremely rare. As such, the extent to which distinct co-occurring taxa experience similar long-term patterns in species richness and compositional change (e.g., when exposed to a changing environment) is not clear. Using data from a diverse ecological community surveyed in the 1930s and resurveyed in the 2010s, we investigated whether local plant and insect assemblages displayed cross-taxon congruence-that is, spatiotemporal correlation in species richness and compositional change-across six co-occurring taxa: vascular plants, non-vascular plants, grasshoppers and crickets (Orthoptera), ants (Hymenoptera: Formicinae), hoverflies (Diptera: Syrphidae), and dragonflies and damselflies (Odonata). All taxa exhibited high levels of turnover across the ca. 80-year time period. Despite minimal observed changes at the level of the whole study system, species richness displayed widespread cross-taxon congruence (i.e., correlated temporal change) across local assemblages within the study system. Hierarchical logistic regression models suggest a role for shared responses to environmental change underlying cross-taxon correlations and highlight stronger correlations between vascular plants and their direct consumers, suggesting a possible role for biotic interactions between these groups. These results provide an illustration of cross-taxon congruence in biodiversity change using data unique in its combination of temporal and taxonomic scope, and highlight the potential for cascading and comparable effects of environmental change (abiotic and biotic) on co-occurring plant and insect communities. However, analyses of historical resurveys based on currently available data come with inherent uncertainties. As such, this study highlights a need for well-designed experiments, and monitoring programs incorporating co-occurring taxa, to determine the underlying mechanisms and prevalence of congruent biodiversity change as anthropogenic environmental change accelerates apace.

Model ensembles of ecosystem services fill global certainty and capacity gaps.

Sustaining ecosystem services (ES) critical to human well-being is hindered by many practitioners lacking access to ES models ("the capacity gap") or knowledge of the accuracy of available models ("the certainty gap"), especially in the world's poorer regions. We developed ensembles of multiple models at an unprecedented global scale for five ES of high policy relevance. Ensembles were 2 to 14% more accurate than individual models. Ensemble accuracy was not correlated with proxies for research capacity, indicating that accuracy is distributed equitably across the globe and that countries less able to research ES suffer no accuracy penalty. By making these ES ensembles and associated accuracy estimates freely available, we provide globally consistent ES information that can support policy and decision-making in regions with low data availability or low capacity for implementing complex ES models. Thus, we hope to reduce the capacity and certainty gaps impeding local- to global-scale movement toward ES sustainability.

Could green infrastructure supplement ecosystem service provision from semi-natural grasslands?

Ancient semi-natural grasslands in Europe are important for ecosystem service (ES) provision. Often, the surrounding matrix contains 'Grassland Green Infrastructure' (GGI) that contain grassland species which have the potential to supplement grassland ES provision across the landscape. Here we investigate the potential for GGI to deliver a set of complementary ES, driven by plant composition.We surveyed 36 landscapes across three European countries comprising core grasslands and their surrounding GGI. We calculated community-level values of plant species characteristics to provide indicators for four ES: nature conservation value, pollination, carbon storage and aesthetic appeal.Inferred ES delivery for GGI was substantially lower than in core grasslands for conservation, pollination and aesthetic appeal indicators, but not for carbon storage. These differences were driven by the GGI having 17% fewer plant species, and compositional differences, with 61% of species unique to the core grasslands. In addition, connectivity to the core, the amount of GGI and inferred seed dispersal distances by livestock, were strongly positively correlated with conservation value, pollination and aesthetic indicators. All ES indicators showed similar responses to the GGI spatial structure and distance to the core, suggesting robust effects of these drivers on ES. We projected that improved landscape-wide delivery of nature conservation value and pollination could be achieved through targeted GGI management. Reductions in the distances seeds would need to disperse, more GGI, along with a diversification of the GGI elements, were predicted to enhance service credits.We conclude that for vegetation-related ES, species surveys can be employed to assess potential ES delivery. Creating and enhancing GGI is a useful landscape management strategy to supplement the ES delivered by ancient grasslands.

Urbanization affects spatial variation and species similarity of bird diversity distribution.

Although cities are human-dominated systems, they provide habitat for many other species. Because of the lack of long-term observation data, it is challenging to assess the impacts of rapid urbanization on biodiversity in Global South countries. Using multisource data, we provided the first analysis of the impacts of urbanization on bird distribution at the continental scale and found that the distributional hot spots of threatened birds overlapped greatly with urbanized areas, with only 3.90% of the threatened birds' preferred land cover type in urban built-up areas. Bird ranges are being reshaped differently because of their different adaptations to urbanization. While green infrastructure can improve local bird diversity, the homogeneous urban environment also leads to species compositions being more similar across regions. More attention should be paid to narrow-range species for the formulation of biodiversity conservation strategies, and conservation actions should be further coordinated among cities from a global perspective.

Environmental morphing enables informed dispersal of the dandelion diaspore.

Animal migration is highly sensitised to environmental cues, but plant dispersal is considered largely passive. The common dandelion, Taraxacum officinale, bears an intricate haired pappus facilitating flight. The pappus enables the formation of a separated vortex ring during flight; however, the pappus structure is not static but reversibly changes shape by closing in response to moisture. We hypothesised that this leads to changed dispersal properties in response to environmental conditions. Using wind tunnel experiments for flow visualisation, particle image velocimetry, and flight tests, we characterised the fluid mechanics effects of the pappus morphing. We also modelled dispersal to understand the impact of pappus morphing on diaspore distribution. Pappus morphing dramatically alters the fluid mechanics of diaspore flight. We found that when the pappus closes in moist conditions, the drag coefficient decreases and thus the falling velocity is greatly increased. Detachment of diaspores from the parent plant also substantially decreases. The change in detachment when the pappus closes increases dispersal distances by reducing diaspore release when wind speeds are low. We propose that moisture-dependent pappus-morphing is a form of informed dispersal allowing rapid responses to changing conditions.

Improving quantitative synthesis to achieve generality in ecology.

Synthesis of primary ecological data is often assumed to achieve a notion of 'generality', through the quantification of overall effect sizes and consistency among studies, and has become a dominant research approach in ecology. Unfortunately, ecologists rarely define either the generality of their findings, their estimand (the target of estimation) or the population of interest. Given that generality is fundamental to science, and the urgent need for scientific understanding to curb global scale ecological breakdown, loose usage of the term 'generality' is problematic. In other disciplines, generality is defined as comprising both generalizability-extending an inference about an estimand from the sample to the population-and transferability-the validity of estimand predictions in a different sampling unit or population. We review current practice in ecological synthesis and demonstrate that, when researchers fail to define the assumptions underpinning generalizations and transfers of effect sizes, generality often misses its target. We provide guidance for communicating nuanced inferences and maximizing the impact of syntheses both within and beyond academia. We propose pathways to generality applicable to ecological syntheses, including the development of quantitative and qualitative criteria with which to license the transfer of estimands from both primary and synthetic studies.

Slow development of woodland vegetation and bird communities during 33 years of passive rewilding in open farmland.

Passive rewilding is a potential tool for expanding woodland cover and restoring biodiversity by abandoning land management and allowing natural vegetation succession to occur. Land can be abandoned to passive rewilding deliberately or due to socio-economic change. Despite abandonment being a major driver of land use change, few have studied the long-term outcomes for vegetation and biodiversity in Western Europe. Studies are also biased towards sites that are close to seed sources and favourable to woodland colonisation. In this case-study, we reconstruct a time series of passive rewilding over 33 years on 25 ha of former farmland that had been subject to soil tipping, far from woodland seed sources. Natural colonisation by shrubs and trees was surveyed at three points during the time series, using field mapping and lidar. Breeding birds were surveyed at three time points, and compared with surveys from nearby farmland. Results showed that natural colonisation of woody vegetation was slow, with open grassland dominating the old fields for two decades, and small wetlands developing spontaneously. After 33 years, thorny shrub thickets covered 53% of the site and former hedgerows became subsumed or degraded, but trees remained scarce. However, the resulting habitat mosaic of shrubland, grassland and wetland supported a locally distinctive bird community. Farmland bird species declined as passive rewilding progressed, but this was countered by relatively more wetland birds and an increase in woodland birds, particularly songbirds, compared to nearby farmland. Alongside biodiversity benefits, shrubland establishment by passive rewilding could potentially provide ecosystem services via abundant blossom resources for pollinators, and recreation and berry-gathering opportunities for people. Although closed-canopy woodland remained a distant prospect even after 33 years, the habitat mosaic arising from passive rewilding could be considered a valuable outcome, which could contribute to nature recovery and provision of ecosystem services.

Understanding the accuracy of modelled changes in freshwater provision over time.

Accurate modelling of changes in freshwater supplies is critical in an era of increasing human demand, and changes in land use and climate. However, there are concerns that current landscape-scale models do not sufficiently capture catchment-level changes, whilst large-scale comparisons of empirical and simulated water yield changes are lacking. Here we modelled annual water yield in two time periods (1: 1985-1994 and 2: 2008-2017) across 81 catchments in England and validated against empirical data. Our objectives were to i) investigate whether modelling absolute or relative change in water yield is more accurate and ii) determine which predictors have the greatest impact on model accuracy. We used the Integrated Valuation of Ecosystem Services and Tradeoffs (InVEST) Annual Water Yield model. In this study, absolute values refer to volumetric units of million cubic metres per year (Mm3/y), either at the catchment or hectare level. Modelled annual yields showed high accuracy as indicated by the low Mean Absolute Deviation (MAD, based on normalised data, 0 is high and 1 is low accuracy) at the catchment (1: 0.013 ± 0.019, 2: 0.012 ± 0.020) and hectare scales (1: 0.03 ± 0.030, 2: 0.030 ± 0.025). But accuracy of modelled absolute change in water yield showed a more moderate fit on both the catchment (MAD = 0.055 ± 0.065) and hectare (MAD = 0.105 ± 0.089) scales. Relative change had lower accuracy (MAD = 0.189 ± 0.135). Anthropogenic modifications to the hydrological system, including water abstraction contributed significantly to the inaccuracy of change values at the catchment and hectare scales. Quantification of changes in freshwater provision can be more accurately articulated using absolute values rather than using relative values. Absolute values can provide clearer guidance for mitigation measures related to human consumption. Accuracy of modelled change is related to different aspects of human consumption, suggesting anthropogenic impacts are critically important to consider when modelling water yield.

The current and future uses of machine learning in ecosystem service research.

Machine learning (ML) expands traditional data analysis and presents a range of opportunities in ecosystem service (ES) research, offering rapid processing of 'big data' and enabling significant advances in data description and predictive modelling. Descriptive ML techniques group data with little or no prior domain specific assumptions; they can generate hypotheses and automatically sort data prior to other analyses. Predictive ML techniques allow for the predictive modelling of highly non-linear systems where casual mechanisms are poorly understood, as is often the case for ES. We conducted a review to explore how ML is used in ES research and to identify and quantify trends in the different ML approaches that are used. We reviewed 308 peer-reviewed publications and identified that ES studies implemented machine learning techniques in data description (64%; n = 308) and predictive modelling (44%), with some papers containing both categories. Classification and Regression Trees were the most popular techniques (60%), but unsupervised learning techniques were also used for descriptive tasks such as clustering to group or split data without prior assumptions (19%). Whilst there are examples of ES publications that apply ML with rigour, many studies do not have robust or repeatable methods. Some studies fail to report model settings (43%) or software used (28%), and many studies do not report carrying out any form of model hyperparameter tuning (67%) or test model generalisability (59%). Whilst studies use ML to analyse very large and complex datasets, ES research is generally not taking full advantage of the capacity of ML to model big data (1138 medium number of data points; 13 median quantity of variables). There is great further opportunity to utilise ML in ES research, to make better use of big data and to develop detailed modelling of spatial-temporal dynamics that meet stakeholder demands.

Long-term woodland restoration on lowland farmland through passive rewilding.

Natural succession of vegetation on abandoned farmland provides opportunities for passive rewilding to re-establish native woodlands, but in Western Europe the patterns and outcomes of vegetation colonisation are poorly known. We combine time series of field surveys and remote sensing (lidar and photogrammetry) to study woodland development on two farmland fields in England over 24 and 59 years respectively: the New Wilderness (2.1 ha) abandoned in 1996, and the Old Wilderness (3.9 ha) abandoned in 1961, both adjacent to ancient woodland. Woody vegetation colonisation of the New Wilderness was rapid, with 86% vegetation cover averaging 2.9 m tall after 23 years post-abandonment. The Old Wilderness had 100% woody cover averaging 13.1 m tall after 53 years, with an overstorey tree-canopy (≥ 8 m tall) covering 91%. By this stage, the structural characteristics of the Old Wilderness were approaching those of neighbouring ancient woodlands. The woody species composition of both Wildernesses differed from ancient woodland, being dominated by animal-dispersed pedunculate oak Quercus robur and berry-bearing shrubs. Tree colonisation was spatially clustered, with wind-dispersed common ash Fraxinus excelsior mostly occurring near seed sources in adjacent woodland and hedgerows, and clusters of oaks probably resulting from acorn hoarding by birds and rodents. After 24 years the density of live trees in the New Wilderness was 132/ha (57% oak), with 390/ha (52% oak) in the Old Wilderness after 59 years; deadwood accounted for 8% of tree stems in the former and 14% in the latter. Passive rewilding of these 'Wilderness' sites shows that closed-canopy woodland readily re-established on abandoned farmland close to existing woodland, it was resilient to the presence of herbivores and variable weather, and approached the height structure of older woods within approximately 50 years. This study provides valuable long-term reference data in temperate Europe, helping to inform predictions of the potential outcomes of widespread abandonment of agricultural land in this region.

Spatial extent of road pollution: A national analysis.

Roads form vast, pervasive and growing networks across the Earth, causing negative environmental impacts that spill out into a 'road-effect zone'. Previous research has estimated the regional and global extent of these zones using arbitrary distances, ignoring the spatial distribution and distance-dependent attenuation of different forms of road environmental impact. With Great Britain as a study area, we used mapping of roads and realistic estimates of how pollution levels decay with distance to project the spatial distribution of road pollution. We found that 25% of land was less than 79 m from a road, 50% of land was less than 216 m and 75% of land was less than 527 m. Roadless areas were scarce, and confined almost exclusively to the uplands (mean elevation 391 m), with only ca 12% of land in Great Britain more than 1 km from roads and <4% of land more than 2.5 km from roads. Using light, noise, heavy metals, NO2, and particulate matter PM2.5 and PM10 as examples, we estimate that roads have a zone of influence that extends across >70% of the land area. Potentially less than 6% of land escapes any impact, resulting in nearly ubiquitously elevated pollution levels. Generalising from this, we find that, whilst the greatest levels of road pollution are relatively localised around the busiest roads, low levels of road pollution (which may be ecologically significant) are pervasive. Our findings demonstrate the importance of incorporating greater realism into road-effect zones and considering the ubiquity of road pollution in global environmental issues. We used Great Britain as a study area, but the findings likely apply to other densely populated regions at present, and to many additional regions in the future due to the predicted rapid expansion of the global road network.

Ecological restoration of agricultural land can improve its contribution to economic development.

Given the negative environmental impacts of intensive agriculture, there is an urgent need to reduce the impact of food production on biodiversity. Ecological restoration of farmland could potentially contribute to this goal. While the positive impacts of ecological restoration on biodiversity are well established, less evidence is available regarding impacts on economic development and employment. Potentially, prospects for economic development could be enhanced by ecological restoration though increased provision of ecosystem services, on which some economic activity depends. Here we examined this issue through the development of contrasting land use scenarios for the county of Dorset, southern England. Two scenarios of future agricultural expansion were compared with two scenarios of landscape-scale ecological restoration and the current situation. Impacts on provision of multiple ecosystem services (ES) were explored using InVEST models and proxy values for different land cover types. Impacts on economic employment were examined using an economic input-output model, which was adjusted for variation in ES flows using empirically determined ES dependency values for different economic sectors. Using the unadjusted input-output model, the scenarios had only a slight economic impact (≤ 0.3% Gross Value Added, GVA). Conversely, when the input-output model was adjusted to take account of ES flows, GVA increased by up to 5.4% in the restoration scenarios, whereas under the scenario with greatest agricultural expansion, GVA was reduced by -4.5%. Similarly, employment increased by up to 6.7% following restoration, compared to declines of up to -5.6% following maximum agricultural expansion. These results show that the economic contribution of rural land is far greater than that attributable to agricultural production alone. Landscape-scale restoration of agricultural land can potentially increase the contribution of farmland to economic development and employment, by increasing flows of multiple ES to the many economic sectors that depend on them.

Enriching the Shared Socioeconomic Pathways to co-create consistent multi-sector scenarios for the UK.

As the pressure to take action against global warming is growing in urgency, scenarios that incorporate multiple social, economic and environmental drivers become increasingly critical to support governments and other stakeholders in planning climate change mitigation or adaptation actions. This has led to the recent explosion of future scenario analyses at multiple scales, further accelerated since the development of the Intergovernmental Panel on Climate Change (IPCC) research community Shared Socioeconomic Pathways (SSPs) and Representative Concentration Pathways (RCPs). While RCPs have been widely applied to climate models to produce climate scenarios at multiple scales for investigating climate change impacts, adaptation and vulnerabilities (CCIAV), SSPs are only recently being scaled for different geographical and sectoral applications. This is seen in the UK where significant investment has produced the RCP-based UK Climate Projections (UKCP18), but no equivalent UK version of the SSPs exists. We address this need by developing a set of multi-driver qualitative and quantitative UK-SSPs, following a state-of-the-art scenario methodology that integrates national stakeholder knowledge on locally-relevant drivers and indicators with higher level information from European and global SSPs. This was achieved through an intensive participatory process that facilitated the combination of bottom-up and top-down approaches to develop a set of UK-specific SSPs that are locally comprehensive, yet consistent with the global and European SSPs. The resulting scenarios balance the importance of consistency and legitimacy, demonstrating that divergence is not necessarily the result of inconsistency, nor comes as a choice to contextualise narratives at the appropriate scale.

Spatial variation in biodiversity loss across China under multiple environmental stressors.

Biodiversity is essential for the maintenance of ecosystem health and delivery of the Sustainable Development Goals. However, the drivers of biodiversity loss and the spatial variation in their impacts are poorly understood. Here, we explore the spatial-temporal distributions of threatened and declining ("biodiversity-loss") species and find that these species are affected by multiple stressors, with climate and human activities being the fundamental shaping forces. There has been large spatial variation in the distribution of threatened species over China's provinces, with the biodiversity of Gansu, Guangdong, Hainan, and Shaanxi provinces severely reduced. With increasing urbanization and industrialization, the expansion of construction and worsening pollution has led to habitat retreat or degradation, and high proportions of amphibians, mammals, and reptiles are threatened. Because distributions of species and stressors vary widely across different climate zones and geographical areas, specific policies and measures are needed for preventing biodiversity loss in different regions.

Ensembles of ecosystem service models can improve accuracy and indicate uncertainty.

Many ecosystem services (ES) models exist to support sustainable development decisions. However, most ES studies use only a single modelling framework and, because of a lack of validation data, rarely assess model accuracy for the study area. In line with other research themes which have high model uncertainty, such as climate change, ensembles of ES models may better serve decision-makers by providing more robust and accurate estimates, as well as provide indications of uncertainty when validation data are not available. To illustrate the benefits of an ensemble approach, we highlight the variation between alternative models, demonstrating that there are large geographic regions where decisions based on individual models are not robust. We test if ensembles are more accurate by comparing the ensemble accuracy of multiple models for six ES against validation data across sub-Saharan Africa with the accuracy of individual models. We find that ensembles are better predictors of ES, being 5.0-6.1% more accurate than individual models. We also find that the uncertainty (i.e. variation among constituent models) of the model ensemble is negatively correlated with accuracy and so can be used as a proxy for accuracy when validation is not possible (e.g. in data-deficient areas or when developing scenarios). Since ensembles are more robust, accurate and convey uncertainty, we recommend that ensemble modelling should be more widely implemented within ES science to better support policy choices and implementation.

Human-mediated dispersal and disturbance shape the metapopulation dynamics of a long-lived herb.

As anthropogenic impacts on the natural world escalate, there is increasing interest in the role of humans in dispersing seeds. But the consequences of this Human-Mediated Dispersal (HMD) on plant spatial dynamics are little studied. In this paper, we ask how secondary dispersal by HMD affects the dynamics of a natural plant metapopulation. In addition to dispersal between patches, we suggest within-patch processes can be critical. To address this, we assess how variation in local population dynamics, caused by small-scale disturbances, affects metapopulation size. We created an empirically based model with stochastic population dynamics and dispersal among patches, which represented a real-world, cliff-top metapopulation of wild cabbage Brassica oleracea. We collected demographic data from multiple populations by tagging plants over eight years. We assessed seed survival, and establishment and survival of seedlings in intact vegetation vs. small disturbances. We modeled primary dispersal by wind using field data and used experimental data on secondary HMD by hikers. We monitored occupancy patterns over a 14-yr period in the real metapopulation. Disturbance had large effects on local population growth rates, by increasing seedling establishment and survival. This meant that the modeled metapopulation grew in size only when the area disturbed in each patch was above 35%. In these growing metapopulations, although only 0.2% of seeds underwent HMD, this greatly enhanced metapopulation growth rates. Similarly, HMD allowed more colonizations in declining metapopulations under low disturbance, and this slowed the rate of decline. The real metapopulation showed patterns of varying patch occupancy over the survey years, which were related to habitat quality, but also positively to human activity along the cliffs, hinting at beneficial effects of humans. These findings illustrate that realistic changes to dispersal or demography, specifically by humans, can have fundamental effects on the viability of a species at the landscape scale.