Ecology

Keith, D. A. et al. Nature https://doi.org/10.1038/s41586-022-05318-4 (2022).Article 

Google Scholar 
Domesday Book, or, The Great Survey of England of William the Conqueror A.D. MLXXXVI (Ordnance Survey Office, 1862).McMahon, G. et al. Environ. Manage. 28, 293–316 (2001).PubMed 
Article 

Google Scholar 
Spalding, M. D. et al. BioScience 57, 573–583 (2007).Article 

Google Scholar 
Holdridge, L. R. Science 105, 367–368 (1947).PubMed 
Article 

Google Scholar 
Köppen, W. in Handbuch der Klimatologie (eds Köppen, W. & Geiger, G. C.) 1–44 (Gebrüder Borntraeger, 1936).
Google Scholar 
Whittaker, R. H. Communities and Ecosystems (Macmillan, 1975).
Google Scholar 
Keddy, P. A. Trends Ecol. Evol. 9, 231–234 (1994).PubMed 
Article 

Google Scholar 
United Nations. Convention on Biological Diversity (UN, 1992).
Google Scholar 
MacArthur, R. H. Geographical Ecology: Patterns in the Distribution of Species (Princeton Univ. Press, 1972).
Google Scholar 
Schoener, T. W. in Community Ecology (eds Diamond, J. D. & Case, T. ) 467–479 (Harper & Row, 1986).
Google Scholar 
Winemiller, K. O., Fitzgerald, D. B., Bower, L. M. & Pianka, E. R. Ecol. Lett. 18, 737–751 (2015).PubMed 
Article 

Google Scholar  More

We developed the IUCN Global Ecosystem Typology in the following sequence of steps: design criteria; hierarchical structure and definition of levels; generic ecosystem assembly model; top-down classification of the upper hierarchical levels; iterative circumscription of the units and ecosystem-specific adaptations of the assembly model; full description of the units; and map compilation. Some iteration proved necessary, as the description and review process sometimes revealed a need for circumscribing additional units.Design criteria and other typologiesUnder the auspices of the IUCN Commission on Ecosystem Management, we developed six design principles to guide the development of a typology that would meet the needs for global ecosystem reporting, risk assessment, natural capital accounting and ecosystem management: (1) representation of ecological processes and ecosystem functions; (2) representation of biota; (3) conceptual consistency throughout the biosphere; (4) scalable structure; (5) spatially explicit units; and (6) parsimony and utility (see Supplementary Table 1.1 and Supplementary Information, Appendix 1 for definitions and rationale).We assessed 23 existing ecological classifications with global coverage of terrestrial, freshwater, and/or marine environments against these principles to determine their fitness for IUCN’s purpose (Supplementary Information, Appendix 1). These include general classifications of land, water or bioclimate, as well as classifications of units that conform with the definition of ecosystems adopted in the United Nations Convention on Biological Diversity45 or an equivalent definition in the IUCN Red List of Ecosystems30. We reviewed documentation on methods of derivation, descriptions of classification units and maps to assess each classification against the six design principles (Supplementary Table 1.2 for details).Typology structure and ecosystem assemblyWe developed the structure of the Global Ecosystem Typology and the generic ecosystem assembly model at a workshop attended by 48 terrestrial, freshwater and marine ecosystem experts at Kings College London, UK, in May 2017. Participants agreed that a hierarchical structure would provide an effective framework for integrating ecological processes and functional properties (Supplementary Table 1.1, design principle 1), and biotic composition (principle 2) into the typology, while also meeting the requirement for scalability (principle 4). Although neither function nor composition were intended to take primacy within the typology, we reasoned that a hierarchy representing functional features in the upper levels is likely to support generalizations and predictions by leveraging evolutionary convergence13. By contrast, a typology reflecting compositional similarities in its upperlevels is less likely to be stable owing to dynamism of species assemblages and evolving knowledge on species taxonomy and distributions. Furthermore, representation of compositional relationships at a global scale would require many more units in upper levels, and possibly more hierarchical levels. Therefore, we concluded that a hierarchical structure recognizing compositional variants at lower levels within broad functionally based groupings at upper levels would be more parsimonious and robust (principle 6) than one representing composition at upper levels and functions at lower levels.Workshop participants initially agreed that three hierarchical levels for ecosystem function and three levels for biotic composition could be sufficient to represent global variation across the whole biosphere. Participants developed the concepts of these levels into formal definitions (Supplementary Table 3.1), which were reviewed and refined during the development process.To ensure conceptual consistency of the typology and its units throughout the biosphere (principle 3), we drew from community assembly theory to develop a generic model of ecosystem assembly. The traditional community assembly model incorporates three types of filters (dispersal, the abiotic environment and biotic interactions) that determine which biota from a larger pool of potential colonists can occupy and persist in an area13. We extended this model to ecosystems by: (1) defining three groups of abiotic filters (resources, ambient environment and disturbance regimes) and two groups of biotic filters (biotic interactions and human activity); (2) incorporating evolutionary processes that shape characteristic biotic properties of ecosystems over time; (3) defining the outcomes of filtering and evolution in terms of all ecosystem properties including both ecosystem-level functions and species-level traits, rather than only in terms of species traits and composition; and (4) incorporating interactions and feedbacks among filters and selection agents and ecosystem properties to elucidate hypotheses about processes that influence temporal and spatial variability in the properties of ecosystems and their component biota. In community assembly, only a small number of filters are likely to be important in any given habitat13. In keeping with this proposition, we used the generic model to identify biological and physical features that distinguish functionally different groups of ecosystems from one another by focusing on different ecological drivers that come to the fore in structuring their assembly and shaping their properties.Hierarchical levelsThe top level of classification (Fig. 2 and Extended Data Tables 1–4) defines five core realms of the biosphere based on contrasting media that reflect ecological processes and functional properties: terrestrial; freshwaters and inland saline waters (hereafter freshwater); marine; subterranean; and atmospheric. Biome gradient concepts25 highlight continuous variation in ecosystem properties, which is represented in the typology by transitional realms that mark the interfaces between the five core realms (for example, floodplains (terrestrial–freshwater), estuaries (freshwater–marine), and so on). In Supplementary Information, Appendix 3 (pages 3–16) and Supplementary Table 3.1, we describe the five core realms and review the hypothesized assembly filters and ecosystem properties that distinguish different groups within them. The atmospheric realm is included for comprehensive coverage, but we deferred resolution of its lower levels because its biota is poorly understood, sparse, itinerant and represented mainly by dispersive life stages46.Functional biomes (level 2) are components of the biosphere united by one or more major assembly processes that shape key ecosystem functions and ecological processes, irrespective of taxonomic identity (Supplementary Information, Appendix 3, page 17). Our interpretation aligns broadly with ‘functional biomes’ described elsewhere24,25,47, extended here to reflect dominant assembly filters and processes across all realms, rather than the more restricted basis of climate-vegetation relationships that traditionally underpin biome definition on land. Hence, the 25 functional biomes (Supplementary Information, Appendix 4, pages 52–186 and https://global-ecosystems.org/) include some ‘traditional’ terrestrial biomes47, as well as lentic and lotic freshwater systems, pelagic and benthic marine systems, and anthropogenic functional biomes assembled and usually maintained by human activity48.Level 3 of the typology defines 110 ecosystem functional groups described with illustrated profiles in Supplementary Information, Appendix 4 (pages 52–186) and at https://global-ecosystems.org/. These are key units for generalization and prediction, because they include ecosystem types with convergent ecosystem properties shaped by the dominance of a common set of drivers (Supplementary Information, Appendix 3, pages 17–19). Ecosystem functional groups are differentiated along environmental gradients that define spatial and temporal variation in ecological drivers (Figs. 2 and 3 and Supplementary Figs. 3.2 and 3.4). For example, depth gradients of light and nutrients differentiate functional groups in pelagic ocean waters (Fig. 3c and Extended Data Table 4), influencing assembly directly and indirectly through predation. Resource gradients defined by flow regimes (influenced by catchment precipitation and evapotranspiration) and water chemistry, modulated by environmental gradients in temperature and geomorphology, differentiate functional groups of freshwater ecosystems25 (Fig. 3b and Extended Data Table 3). Terrestrial functional groups are distinguished primarily by gradients in water and nutrient availability and by temperature and seasonality (Fig. 3a and Extended Data Table 1), which mediate uptake of those resources and regulate competitive dominance and productivity of autotrophs. Disturbance regimes, notably fire, are important global drivers in assembly of some terrestrial ecosystem functional groups49.Three lower levels of the typology distinguish functionally similar ecosystems based on biotic composition. Our focus in this paper is on global functional relationships of ecosystems represented in the upper three levels of the typology, but the lower levels (Supplementary Information, Appendix 3, pages 19 and 20) are crucial for representing the biota in the typology, and facilitate the scaling up of information from established local-scale typologies that support decisions where most conservation action takes place. These lower levels are being developed progressively through two contrasting approaches with different trade-offs, strengths and weaknesses. First, level 4 units (regional ecosystem subgroups) are ecoregional expressions of ecosystem functional groups developed from the top-down by subdivisions based on biogeographic boundaries (for example, in ref. 50) that serve as simple and accessible proxies for biodiversity patterns51. Second, level 5 units (global ecosystem types) are also regional expressions of ecosystem functional groups, but unlike level 4 units they are explicitly linked to local information sources by bottom-up aggregation52 and rationalization of level 6 units from established subglobal ecological classifications. Subglobal classifications, such as those for different countries (see examples for Chile and Myanmar in Supplementary Tables 3.3 and 3.4), are often developed independently of one another, and thus may involve inconsistencies in methods and thematic resolution of units (that is, broadly defined or finely split). Aggregation of level 6 units to broader units at level 5 based on compositional resemblance is necessary to address inconsistencies among different subglobal classifications and produce compositionally distinctive units suitable for global or regional synthesis.Integrating local classifications into the global typology, rather than replacing them, exploits considerable efforts and investments to produce existing classifications, already developed with local expertise, accuracy and precision. By placing national and regional ecosystems into a global context, this integration also promotes local ownership of information to support local action and decisions, which are critical to ecosystem conservation and management outcomes (Supplementary Information, Appendix 3, page 20). These benefits of bottom-up approaches come at the cost of inevitable inconsistencies among independently developed classifications from different regions, a limitation avoided in the top-down approach applied to level 4.Circumscribing upper-level unitsWe formed specialist working groups (terrestrial/subterranean, freshwater and marine) to develop descriptions of the units within the upper levels of the hierarchy, subdividing realms into functional biomes, and biomes into ecosystem functional groups. We used definitions of the hierarchical levels (Supplementary Table 3.1) and the conceptual model of ecosystem assembly (Fig. 1) to maintain consistency in defining the units at each level during iterative discussions within and between the working groups.Working groups agreed on preliminary lists of functional biomes and ecosystem functional groups by considering variation in major drivers along ecological gradients (Figs. 2 and 3 and Supplementary Figs. 3.2 and 3.4) based on published literature, direct experience and expertise of working group members, and consultation with colleagues in their respective research networks. After the workshop, working groups sought recent global reviews of the candidate units and recent case studies of exemplars to shape descriptions of the major groups of ecosystem drivers and properties for each unit. Circumscriptions and descriptions of the units were reviewed and revised iteratively to ensure clear distinctions among units, with a total of 206 reviews of descriptive profiles undertaken by 60 specialists, a mean of 2.4 reviews per profile (Supplementary Table 5.1). The working groups concurrently adapted the generic model of ecosystem assembly (Fig. 1) to represent working hypotheses on salient drivers and ecosystem properties for each ecosystem functional group.Incorporating human influenceVery few of the ecological typologies reviewed in Supplementary Information, Appendix 1 integrate anthropogenic ecosystems in their classificatory frameworks. Anthropogenic influences create challenges for ecosystem classification, as they may modify defining features of ecosystems to a degree that varies from negligible to major transformation across different locations and times. We addressed this problem by distinguishing transformative outcomes of human activity at levels 2 and 3 of the typology from lesser human influences that may be represented either at levels 5 and 6, or through measurements of ecosystem integrity or condition that reflect divergence from reference states arising from human activity.Anthropogenic ecosystems grouped within levels 2 and 3 were thus defined as those created and sustained by intensive human activities, or arising from extensive modification of natural ecosystems such that they function very differently. These activities are ultimately driven by socio-economic and cultural-spiritual processes that operate across local to global scales of human organization. In many agricultural and aquacultural systems and some others, cessation of those activities may lead to transformation into ecosystem types with qualitatively different properties and organizational processes (see refs. 53,54 for cropland and urban examples, respectively). Indices such as human appropriation of net primary productivity55, combined with land-use maps56, offer useful insights into the distribution of some anthropogenic ecosystems, but further development of indices is needed to adequately represent others, particularly in marine, and freshwater environments. Beyond land-use classification and mapping approaches (Supplementary Information, Appendix 1, page 6), a more comprehensive elaboration of the intensity of human influence underpinning the diverse range of anthropogenic ecosystems requires a multidimensional framework incorporating land-use inputs, outputs, their interactions, legacies of earlier activity and changes in system properties17.Where less intense human activities occur within non-anthropogenic ecosystem types, we focused descriptions on low-impact reference states. Therefore, human activities are not shown as drivers in the assembly models for non-anthropogenic ecosystem groups, even though they may have important influences on the contemporary ecosystem distribution. This approach enables the degree and nature of human influence to be described and measured against these reference states using assessment methods such as the Red List of Ecosystems protocol30, with appropriate data on ecosystem change.Indicative distribution mapsFinally, to produce spatially explicit representations of the units at level 3 of the typology (principle 5), we sought published global maps (sources in Supplementary Table 4.1) that were congruent with the concepts of respective ecosystem functional groups. Where several candidate maps were available, we selected maps with the closest conceptual alignment, finest spatial resolution, global coverage, most recent data and longest time series. The purpose of maps for our study was to visualize global distributions. Prior to applications of map data to spatial analysis, we recommend critical review of methods and validation outcomes reported in each data source to ensure fitness for purpose (Supplementary Information, Appendix 4).Extensive searches of published literature and data archives identified high-quality datasets for some ecosystem functional groups (for example, T1.3 Tropical–subtropical montane rainforests; MT1.4 Muddy shorelines; M1.5 Sea ice) and datasets that met some of these requirements for a number of other ecosystem functional groups (see Supplementary Table 4.1 for details). Where evaluations by authors or reviewers identified limitations in available maps, we used global environmental data layers and biogeographic regionalizations as masks to adjust source maps and improve their congruence to the concept of the relevant functional group (for example, F1.2 Permanent lowland rivers). For ecosystem functional groups with no specific global mapping, we used ecoregions50,57,58 as biogeographic templates to identify broad areas of occurrence. We consulted ecoregion descriptions, global and regional reviews, national and regional ecosystem maps, and applied in situ knowledge of participating experts to identify ecoregions that contain occurrences of the relevant ecosystem functional group (for example, T4.4 Temperate woodlands) (see Supplementary Table 4.1 for details). We mapped ecosystem functional groups as major occurrences where they dominated a landscape or seascape matrix and minor occurrences where they were present, but not dominant in landscape–seascape mosaics, or where dominance was uncertain. Although these two categories in combination communicate more information about ecosystem distribution than binary maps, simple spatial overlays using minor occurrences are likely to inflate spatial statistics. The maps are progressively upgraded in new versions of the typology as explicit spatial models are developed and new data sources become available (see ref. 27 for a current archive of spatial data).The classification and descriptive profiles, including maps, for each functional biome and ecosystem functional group underwent extensive consultation, and targeted peer review and revision through a series of four phases described in Supplementary Information, Appendix 5 (pages 2–4). The reviewer comments and revisions from targeted peer review are documented in Supplementary Table 5.1. In all, more than 100 ecosystem specialists have contributed to the development of v2.1 of the typology.LimitationsUneven knowledge of Earth’s biosphere has constrained the delimitation and description of units within the typology. There is a considerable research bias across the full range of Earth’s ecosystems, with few formal research studies evaluating the relative influence of different ecosystem drivers in many of the functional groups, and abiotic assembly filters generally receiving more attention than biotic and dispersal filters. This poses challenges for developing standardized models of assembly for each ecosystem functional group. The models therefore represent working hypotheses, for which available evidence varies from large bodies of published empirical evidence to informal knowledge of ecosystem experts and their extensive research networks. Large numbers of empirical studies exist for some forest functional groups, savannas, temperate heathlands in Mediterranean-type climates, coral reefs, rocky shores, kelp forests, trophic webs in pelagic waters, small permanent freshwater lakes, and others (see references in the respective profiles (Supplementary Information, Appendix 4)). For example, Bond49 reviewed empirical and modelling evidence on the assembly and function of tropical savannas that make up three ecosystem functional groups, showing that they have a large global biophysical envelope that overlaps with tropical dry forests, and that their distribution and dynamics within that envelope is strongly influenced by top-down regulation via biotic filters (large herbivores and their predators) and recurrent disturbance regimes (fires). Despite the development of this critical knowledge base, savannas suffer from an awareness disparity that hinders effective conservation and management59. In other ecosystems, our assembly models rely more heavily on inferences and generalizations of experts drawn from related ecosystems, are more sensitive to interpretations of participating experts, and await empirical testing and adjustment as understanding improves. Empirical tests could examine hypothesized variation in ecosystem properties along gradients within and between ecosystem functional groups and should return incremental improvements on group delineation and description of assembly processes.High-quality maps at suitable resolution are not yet available for the full set of ecosystem functional groups, which limits current readiness for global analysis. The maps most fit for global synthesis are based on remote sensing and environmental predictors that align closely to the concept of their ecosystem functional group, incorporate spatially explicit ground observations and have low rates of omission and commission errors, ‘high’ spatial resolution (that is, rasters of 1 km2 (30 arcsec) or better), and time series of changes. Sixty of the maps currently in our archive27 aligned directly or mostly with the concept of their corresponding ecosystem functional group, while the remainder were based on indirect spatial proxies, and most were derived from polygon data or rasters of 30 arcsec or finer (Supplementary Table 4.1). Maps for 81 functional groups were based either on known records, or on spatial data validated by quantitative assessments of accuracy or efficacy. Therefore, we suggest that maps currently available for 60–80 of the 110 functional groups are potentially suitable for global spatial analysis of ecosystem distributions. Although, a significant advance on broad proxies such as ecoregions, the maps currently available for ecosystem functional groups would benefit from expanded application of recent advances in remote sensing, environmental datasets, spatial modelling and cloud computing to redress inequalities in reliability and resolution. The most urgent priorities for this work are those identified in Supplementary Table 4.1 as relying on indirect proxies for alignment to concept, qualitative evaluation by experts and coarse resolution ( >1 km2) spatial data.Reporting summaryFurther information on research design is available in the Nature Research Reporting Summary linked to this article. More

Description of PTEsThe descriptive statistics of the contents of soil PTEs in the study area were shown in Table 1. From Table 1, the mean contents of As and Ni in the oil-affected soils exceeded their corresponding risk screening values33, which may damage the soil ecological environment and affect crop growth. Compared with the secondary standard of soil environmental quality34, the mean contents of As, Cu and Zn were all lower than their corresponding Grade II standard values, but the mean contents of Cd, Cr, Ni and Pb in the oil-affected soils were 1.07, 7.46, 7.14 and 1.36 times of their standard values. In contrast with the background value of Hubei province35, except Mn, the mean contents of As, Cd, Cr, Cu, Ni, Pb, Zn and Ba in the oil-affected soils all exceeded their background values. Meanwhile, the variation coefficient of Cr (1.41) was greater than 1. In general, the soil Cd concentration in the study area was higher than that around Gudao Town, a typical oil-producing region of the Shengli Oilfield in the Yellow River Delta, China12, and from Yellow River Delta, a traditional oil field in China9, but was lower than that around two crude oil flow stations in the Niger Delta, Nigeria36. The concentrations of other PTEs were higher than the corresponding element concentrations, detected in the soil around Gudao Town, a typical oil-producing region of the Shengli Oilfield in the Yellow River Delta, China12, from Yellow River Delta, a traditional oil field in China9, and around two crude oil flow stations in the Niger Delta, Nigeria36. The above analysis exhibited that PTEs in the oil-affected soils had a certain degree of accumulation and may be affected by human activities.Table 1 Statistical characteristics for potential toxic elements in in the study area (mg·kg−1).Full size tableLevels of PTEs enrichment and pollutionThe EF and PLI of soil PTEs in the study area were calculated to evaluate the pollution degree of soil PTEs. The calculation results of EF and PLI were shown in Fig. 2 and Table S4. From Fig. 2, the mean EF values of PTEs were showed as Pb  > Cr  > Ni  > As  > Cd  > Zn  > Cu  > Ba. The mean EFs of all PTEs were greater than 1. Among them, the average EF of Cu, Zn and Ba was between 1 and 2, which was slightly enriched. And As (2.18) and Cd (2.12) were moderately enriched. In particular, the average EF values of Cr, Ni and Pb were 14.23, 8.69 and 15.45, respectively, reaching a significant enrichment level, and all samples of Cr, Ni and Pb were at moderate or above enrichment, of which 10% of the Cr samples were extreme pollution, 85% of Cr samples, 95% of Ni and 5% of Pb (Table S4) were significantly enriched. These proved that these PTEs were generally enriched in the study area, especially Cr, Ni and Pb.Figure 2The map of enrichment factor and contamination factor of PTEs in the study area.Full size imageExcept Mn, the average CF values of other PTEs were all  > 1 (Fig. 2), indicating that the accumulation of Mn in the study area was relatively light, and there was no obvious Mn pollution. The CF values of all samples of As, Cr, Ni and Pb, 80% of Cd samples, 75% of Cu samples, 30% of Mn samples, 65% of Zn samples and 75% of Ba samples (Table S4) were higher than 1. And the mean CF values of Cr, Ni and Pb were 14.21, 7.58 and 12.73, respectively, certifying that the pollution of Cr, Ni and Pb in the study area was considerably serious. PLI was calculated based on the CF value of PTEs, and the results were shown in Fig. 2. The average value of PLI was 2.62, indicating that the soil PTEs in the study area were seriously polluted.Spatial distribution of soil PTEs in the study areaGeostatistical analysis was utilized to do ordinary Kriging interpolation of the PTEs in the study area, the results were shown in Fig. 3. As shown in Fig. 3, the spatial distribution of As, Cr, Ni, Zn and Ba was relatively consistent, and their hot spots were concentrated in the southeast, northwest, and central and eastern parts of the study area where oil wells were distributed. The spatial distribution of Cr and Ni exhibited that there were large-scale hotspots near the oil wells, and the content of Cr and Ni in these hotspots was much higher than second-level environmental quality standards of China, which proved that the content of soil Cr and Ni was significantly affected by the oil production activities of the oil production plant. There were crude oil leaks in B and C, and the contents of Zn and Ba in the vicinity of these two oil wells were relatively high, indicating that soil Zn and Ba in this area may be affected by the crude oil leakage, resulting in a certain degree of accumulation in the soil. The area with the second highest As content mainly resided in the middle of the study area. According to the survey, the herbicides were sprayed every year around the H oil well in the middle of the study area, indicating that the accumulation of As in the soil was not only related to oil extraction activities, but also to the use of pesticides (contains copper arsenate, sodium arsenate, etc.)10, 14. In addition, the hot spots of spatial distribution of Pb, Cd and Mn were concentrated in the southeast, and Cu was mainly concentrated in the southeast and midwest. As analyzed above, in addition to Mn, the PTEs Pb, Cd and Cu all have a certain degree of accumulation. And the investigation found that there were many petroleum machinery manufacturing plants in the central and eastern part of the study area, therefore, the accumulation of Pb, Cd and Cu in the soil may be related to factors such as petroleum extraction, crude oil leakage and machinery manufacturing. The above analysis indicated that the influence of human activities is evident on the distribution of soil PTEs3, 23.Figure 3spatial distribution map of soil PTEs in the study area.Full size imagePotential ecological risk assessmentThe potential ecological risk assessment model after adjusting the threshold was used to evaluate the PER of the oil production plant. The individual potential ecological risk of PTEs was shown in Table 2. From Table 2, the average ({E}_{r}^{i}) values of PTEs were Cr  > Pb  > Cd  > Ni  > As  > Cu  > Zn  > Mn. The average ({E}_{r}^{i}) values of Cr and Pb were 79.62 and 63.64, respectively, reaching a relatively high level of potential ecological risk; the average ({E}_{r}^{i}) values of Cd and Ni were 55.95 and 37.91, respectively, which were at medium potential ecological risk level; the average ({E}_{r}^{i}) values of other PTEs were all lower than 30, with minor potential ecological risk. Specifically, all samples of Cu, Mn and Zn were at slight potential ecological risk level; 5% of As samples, 80% of Cd, 85% of Cr, 80% of Ni and 100% of Pb (Table S5) were at medium and above potential ecological risk. In particular, the potential ecological risks of 35% of Cd samples, 10% of Cr samples, 5% of Ni samples and 80% of Pb samples (Table S5) were relatively high, 10% Cd samples reached high potential ecological risk level, and 10% Cr samples had extremely high potential ecological risk. In summary, Geostatistical analysis shows that the hotspot distribution of all PTEs in the study area is almost related to the distribution of oil wells. In addition, the hotspot distribution of PTEs may also be related to factors such as agricultural and industrial activities3. The average value of PER in the study area was 265.08, and the proportions of the three risk levels of medium, slightly high and high were 5%, 75% and 20%, respectively (Table S5). It proved that the study area was at a higher potential ecological risk. Among them, the PER values of samples A, B, D, E, F, G, H, I and J (Table 2) were all greater than 280, reaching fairly high ecological risk.Table 2 Single ecological risk index and potential ecological risk of soil PTEs in study area.Full size tableHuman health risk assessmentThe non-carcinogenic risk assessment of As, Cd, Cr, Cu, Mn, Ni, Pb, Zn and Ba in the soils of the study area was carried out, and the assessment results were shown in Table 3. The THI values of children and adults under the three exposure routes of soil PTEs in the study area were 7.31 and 1.03, respectively, and the THI values were all  > 1, which indicated that soil PTEs around the oil production plants posed significant non-carcinogenic health risks to children and adults. The non-carcinogenic hazardous quotient (HQ) of children and adults in Table 3 revealed that the HQ of all PTEs for adults under each exposure route was less than 1, while the HQ of Cr and Pb for children under the oral intake route was greater than 1, which were 4.91 and 1.17, respectively. For HQ with different exposure routes of the same PTE, each soil PTE presented the risk of oral ingestion  > oral and nasal inhalation risk  > skin contact risk. The result was in agreement with the reports14, 37. Therefore, oral intake was the main exposure route of non-carcinogenic risk, and oral intake of Cr and Pb caused serious non-carcinogenic risk to children. Statistical analysis of HI for soil PTEs in the study area showed that the HI values of PTEs for children were significantly higher than those of adults, and the HI values of PTEs in children and adults were all Cr  > Pb  >   > As  > Ni  > Mn  > Ba  > Cu  > Zn  > Cd. Among them, the HI values of all PTEs for adults were less than 1, indicating that the non-carcinogenic risks caused by a single PTE did not have a significant impact on adults; while the HI values of Cr and Pb for children were 4.93 and 1.17 greater than 1, indicating that they have caused serious non-carcinogenic risk to local children. In addition, the HI values of As and Ni for children and the HI values of As, Cr and Pb for adults were all greater than 0.1, which requires attention. In summary, children suffered from significant non-carcinogenic risk, and adults suffered from minor non-carcinogenic risk in the study area; soil Cr and Pb were the most important non-carcinogenic risk factors for children and adults in the study area.Table 3 Non-cancer and cancer risk assessment of adults and children under different exposure routes.Full size tableIn this study, soil As, Cd, Cr, Ni and Pb from the study area were assessed for carcinogenic risk, and the results were shown in Table 3. The TCRI of children and adults under the three exposure routes of these five PTEs were 9.44E−04 and 5.75E−04, respectively, indicating that soil PTEs around the oil production plants have caused serious carcinogenic risk to local children and adults. The CR values of children and adults showed that the CR values of Cr (6.33E−04) and Ni (2.64E−04) for children, and Cr (3.87E−04) and Ni (1.49E−04) for adults were all greater than 10–4. In addition, As, Cr and Cd all presented oral intake risk  > oronasal inhalation risk  > skin contact risk. In conclusion, Cr and Ni caused serious carcinogenic risk for children and adults in the study area, and oral intake was also the primary way of carcinogenic risk. The CRI statistics of adults and children exhibited that the CRI values of all PTEs were lower than those of children. The CRI values of the PTEs in adults and children under the three exposure routes were Cr  > Ni  >   > As  > Pb  >   > Cd. Among them, the CRI values of Cr and Ni in children and adults by oral intake were both greater than 10–4, showing a strong carcinogenic risk. It is noteworthy that the assessment based on total concentrations of PTEs in soil might overestimate potential health risks38. The above analysis revealed that both children and adults in the study area suffered from serious carcinogenic risks, and Cr and Ni were the chiefly carcinogenic risk factors. More

Characteristics of LVD dataset and assembled vOTUsThe land use virome dataset LVD was derived from 2.6 billion paired clean reads of sequences across 50 viromes of 25 samples with five types of land uses (Supplementary Data 2). A total of 6,442,065 contigs ( >1500 bp) were yielded, of which 764,466 (11.8%) contigs were identified as putative viral genomes through VIBRANT. Subsequently, putative false positive viral genomes were removed (see Methods section), and 27,951 and 48,936 bona fide viral genomes were retained from the 25 intracellular VLPs (iVLPs) and 25 extracellular VLPs (eVLPs) viromes, respectively. These genomes were clustered into 25,941 and 45,152 vOTUs for iVLPs and eVLPs viromes, respectively, in which the iVLPs and eVLPs viromes shared 11,467 (19.2%) vOTUs. Subsequently, they were merged and dereplicated, resulting in 59,626 vOTUs (Supplementary Data 3) for the following analysis. A total of 8112 (13.6%) vOTUs genomes were classified as complete, in which the median length of all and circular vOTUs were 25,183 bp and 45,511 bp, respectively (Supplementary Fig. 4).To explore the taxonomic affiliation of vOTUs in family and genus-level, a gene-sharing network consist of 59,626 vOTUs genomes from this study and 3502 reference phage genomes (from NCBI Viral RefSeq version 201) revealed 6009 VCs comprising of 37,224 vOTUs, of which 34,417 vOTUs were from LVD, besides 2794 singletons (2653 from LVD dataset), 16,056 outliers (15,833 from LVD) and 8492 overlaps (8061 from LVD) were detected (Supplementary Data 4). Of these, only 157 VCs contained genomes from both the RefSeq and LVD dataset (1864 viral genomes) (Supplementary Data 4). Most of VCs (1837, 30.4%) included only two members.At the family level, most of vOTUs were classified into Siphoviridae (712 by vConTACT2 and 29,671 (50.9%) by Demovir, tailed dsDNA), Podoviridae (610 by vConTACT2 and 9923 (17.6 %) by Demovir, tailed dsDNA), Myoviridae (485 by vConTACT2 and 5445 (9.9%) by Demovir, tailed dsDNA), Tectiviridae (50 by vConTACT2 and 10 (0.10%) by Demovir, non-tailed dsDNA) (Fig. 1). Besides, the Eukaryotic viruses Herpesviridae (159 by Demovir, 0.26%, dsDNA), Phycodnaviridae (120 (0.20%) by Demovir, dsDNA); the Virophage Family Lavidaviridae (15 (0.03%) by Demovir) were detected as well, but a majority of vOTUs were unclassified in genus-level.Fig. 1: The taxonomic assignment of LVD.Pie charts showing the affiliation of 56,870 vOTUs at family level assigned by script Demovir (a). and the affiliation of 1864 vOTUs at family level assigned by package vConTACT2 (b). Source data are provided in the Source Data file.Full size imageViral community structures differ across land use typesBray–Curtis dissimilarity of viral communities (median 0.9951) showed strong heterogeneity of viral communities among different sites (Fig. 2a). While, the Bray–Curtis dissimilarity (median: 0.5109) between paired viral communities of iVLPs and eVLPs from each site have a significant lower heterogeneity than inter-sites (Wilcox.test, p  0.05; Fig. 2b). Therefore, the paired iVLPs and eVLPs viromes from each site were merged for subsequently viral community analysis.Fig. 2: The macrodiversity of soil viral communities.a Boxplot showing Bray–Curtis dissimilarity of viral communities of intra-sites (between the corresponding community of iVLPs and eVLPs, n = 25) and inter-sites (between different sample sites, n = 300). The minima, maxima, center, bounds of box and whiskers in boxplots from bottom to top represented percentile 0, 10, 25, 50, 75, 90, and 100, respectively, the difference between different zones was tested using the two-sided Wilcox.test, ****p  More

Study site and field data collectionPermanent 1 m2 photoquadrats were established on Heron Reef in 1962/63, using 9 mm diameter mild steel (rebar) pegs, which were replaced over time. From the 1990’s, replacement pegs were stainless steel for greater longevity. Four sites were established, the protected (south) crest, inner flat, exposed (north) crest and exposed pools. Co-ordinates for each site are presented in Table 1, the layout shown in Fig. 2, and sites have been well described previously5,6. At each census, a 1 m2 frame divided into a 5 × 5 grid using string was placed over the pegs, and the quadrat photographed from directly above at low tide. From 1963 until 2003, a 35 mm camera and colour slide film were used. The camera was attached to a tripod affixed to the 1 m2 frame, and captured around 2/3 of the quadrat. The frame (and camera) were then rotated 180 degrees to capture the remainder of the quadrat. After 2003, a hand-held digital camera was used, with the entire quadrat being captured in a single image. Concurrent with each census, mud maps of each quadrat were hand drawn in the field, and all colonies identified in situ by someone with expertise in coral taxonomy.Table 1 Coordinates of the study sites on Heron Island Reef (WGS84).Full size tableFig. 2Quadrat layouts for each of the four sites respectively, noting that the north crest and north ridge have been treated as a single north crest site in previous publications. Underlining indicates original 1962/63 quadrats. Other quadrats were added in or after 2008, as indicated in the text. Contiguous quadrats are pictured bordering each other. Spacing between separate quadrats or groups of quadrats is not shown to scale. Note that up until 2005, NRNW was known as NR. The acronyms in each quadrat represent its name.Full size imageAt the protected (south) crest, a set of six contiguous quadrats were established in 1963 in a 2 × 3 arrangement parallel to the waterline, and about 420 m southeast of the island. This site is exposed at low tide, and was photographed once all water had drained off it. Images of quadrats A, C & E (the shoreward row) from 1963 to 2012 have been fully processed, and the data have been through QA/QC. Data after 2012 exist as images only. These quadrats form the basis of previous analyses1,4,5,6 for this site. Photographs are available for quadrats B, D & F, but apart from 2003–2010, have not been processed. In 2010, an additional two quadrats were established either side of the original six, leading to a 2 × 5 arrangement. Again, only imagery is available for these additional quadrats.At the inner flat, two pairs of contiguous quadrats were established in 1962, 44 m apart, about 70 m south of the island. This site is covered by ~10 cm of water at low tide, so could only be photographed on a still day. Imagery for this site is only available to 2012, after which the marker stakes appear to have been removed in a cleanup of the area. Images for one quadrat in each pair have been processed, but have not been subject to full QA/QC.At the exposed (north) crest main site, a set of four contiguous quadrats was established about 1100 m northeast of the island in 1963. An additional single quadrat (north ridge) was established 326 m to the east. Images from 1963 to 2012 have been fully processed, and the data have been through QA/QC. Data after 2012 exist as images only. In 2005, the single north ridge quadrat was expanded to 4 m2, and in 2008, both subsites were expanded to six quadrats in a 2 × 3 arrangement. These additional quadrats have been digitised up to 2012, but have not been through full QA/QC.The exposed pools are two individual quadrats about 5 m apart about 30 m north of the eastern (north ridge) exposed crest site. These are on the edge of a natural pool, and range from ~5–50 cm deep at low tide, and so could only be photographed on a calm day. Imagery for this site is only available until 2005, after which the marker stakes could not be relocated. Images from 1963 to 1998 have been processed, but have not been through full QA/QC.Retrieval of coral composition data from the photoquadratsProcessing of the images involved scanning the colour slides to produce digital images, and then orthorectifying each image to a 1 m2 basemap in ArcGIS (ESRI Ltd). The corners of the frame, and the holes for the string grid, were used as control points for the orthorectification. For images that originated as colour slides, each half of the quadrat was individually orthorectified to the same basemap, producing a single image of the entire quadrat (see Fig. 3). While contiguous quadrats were orthorectified individually, they were done so against a basemap containing all quadrats in the group, meaning that the resulting images can be easily merged to create a single image of the group. The outlines of all visible coral colonies ( >~1 cm2), and other benthic organisms such as algae and clams, were then digitised in ArcGIS to create a single shapefile for each quadrat for each year. Each colony was represented as an individual feature within the shapefile, and was assigned a unique colony number and species based on the mud maps drawn in the field. Colony numbers were consistent across years, allowing individual colonies to be tracked over time. If a colony underwent fission, the original colony number was retained for each, with the addition of a unique identifier after a decimal point. For example, if colony 35 split in two, the resultant colonies were identified as 35.1 and 35.2. If 35.2 later split again, the resultant colonies were identified as 35.2.1 and 35.2.2. If the colony overlapped the edge of the quadrat, only the area within the quadrat was digitised, and a flag was applied to indicate that only part of the colony was included (edgestatus = 1 in the data). Upon completion of digitisation, ArcGIS was used to calculate the area and perimeter of all colonies. While multiple census were conducted in 1963, 1971 and 1983, only a single census in each year has been processed. There are currently no plans to undertake further digitisation or QA/QC of this data set.Fig. 3Example orthorectified and stitched (prior to 2001) images from the NCNE quadrat, showing the effects of a cyclone that removed all colonies in 1972, and slow recovery over subsequent decades.Full size image More

Bartz, K. K. & Naiman, R. J. Effects of Salmon-Borne nutrients on riparian soils and vegetation in Southwest Alaska. Ecosystems 8, 529–545 (2005).Article 

Google Scholar 
Erskine, P. D. et al. Subantarctic Macquarie Island—a model ecosystem for studying animal-derived nitrogen sources using 15N natural abundance. Oecologia 117, 187–193 (1998).ADS 
PubMed 
Article 

Google Scholar 
Hocking, M. D. & Reimchen, T. E. Salmon species, density and watershed size predict magnitude of marine enrichment in riparian food webs. Oikos 118(9), 1307–1318 (2009).Article 

Google Scholar 
Hocking, M. D. & Reynolds, J. D. Impacts of salmon on riparian plant diversity. Science 331, 1609–1612 (2011).ADS 
CAS 
PubMed 
Article 

Google Scholar 
Hocking, M. D., & Reimchen, T. E. Salmon-derived nitrogen in terrestrial invertebrates from coniferous forests of the Pacific Northwest. BMC Ecol. 2, 4. https://doi.org/10.1186/1472-6785-2-4 (2002).Bilby, R. E., Fransen, B. R. & Bisson, P. A. Incorporation of nitrogen and carbon from spawning coho salmon into the trophic system of small streams: Evidence from stable isotopes. Can. J. Fish Aquat. Sci. 53, 164–173 (1996).Article 

Google Scholar 
Talley, D. M. et al. Research challenges at the land–sea interface. Estuar. Coast. Shelf Sci. 58, 699–702 (2003).ADS 
Article 

Google Scholar 
Mizutani, H. & Wada, E. Nitrogen and carbon isotope ratios in seabird rookeries and their ecological implications. Ecology 69(2), 340–349 (1988).Article 

Google Scholar 
Rowe, J. A., Litton, C. M., Lepczyk, C. A. & Popp, B. N. Impacts of endangered seabirds on nutrient cycling in montane forest ecosystems of Hawai’i. Pac. Sci. 71(4), 495–509 (2017).Article 

Google Scholar 
Sanchez-Pinero, F. & Polis, G. A. Bottom-up dynamics of allochthonous input: Direct and indirect effects of seabirds on islands. Ecology 81(11), 3117–3132 (2000).Article 

Google Scholar 
Wait, D. A., Aubrey, D. P. & Anderson, W. B. Seabird guano influences on desert islands: Soil chemistry and herbaceous species richness and productivity. J. Arid Environ. 60, 681–695 (2005).ADS 
Article 

Google Scholar 
Stapp, P., Polis, G. A. & Pinero, F. S. Stable isotopes reveal strong marine and El Nino effects on island food webs. Nature 401, 467–469 (1999).ADS 
CAS 
Article 

Google Scholar 
Anderson, W. B., Wait, D. A. & Stapp, P. Resources from another place and time: Responses to pulses in a spatially subsidized system. Ecology 89(3), 660–670 (2008).PubMed 
Article 

Google Scholar 
Ellis, J. C. Marine birds on land: A review of plant biomass, species richness, and community composition in seabird colonies. Plant Ecol. 181(2), 227–241 (2005).Article 

Google Scholar 
Fukami, T. et al. Above- and below-ground impacts of introduced predators in seabird-dominated island ecosystems. Ecol. Lett. 9, 1299–1307 (2006).PubMed 
Article 

Google Scholar 
Wootton, J. T. Direct and indirect effects of nutrients on intertidal community structure: Variable consequences of seabird guano. J. Exp. Mar. Biol. Ecol. 151, 139–153 (1991).Article 

Google Scholar 
McCauley, D. J., et al., From wing to wing: the persistence of long ecological interaction chains in less-disturbed ecosystems. Sci. Rep. 2, 409. https://doi.org/10.1038/srep00409 (2012).Young, H. S., McCauley, D. J., Dunbar, R. B. & Dirzo, R. Plants cause ecosystem nutrient depletion via the interruption of bird-derived spatial subsidies. Proc. Natl. Acad. Sci. U.S.A. 107(5), 2072–2077 (2010).ADS 
CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Lindeboom, H. J. The nitrogen pathway in a Penguin rookery. Ecology 65(1), 269–277 (1984).CAS 
Article 

Google Scholar 
Mizutani, H., Kabaya, Y. & Wada, E. Ammonia volatilization and high 15N/14N ratio in a penguin rookery in Antarctica. Geochem. J. 19(6), 323–327 (1985).ADS 
CAS 
Article 

Google Scholar 
Anderson, W. B. & Polis, G. A. Nutrient fluxes from water to land: seabirds affect plant nutrient status on Gulf of California islands. Oecologia 118, 324–332 (1999).ADS 
PubMed 
Article 

Google Scholar 
Polis, G. A. & Hurd, S. D. Linking marine and terrestrial food webs: Allochthonous input from the ocean supports high secondary productivity on small islands and coastal land communities. Am. Nat. 147, 396–423 (1996).Article 

Google Scholar 
Goss, N. S. New and rare birds found breeding on the San Pedro Martir Isle. University of California Press 5, 240–244 (1888).
Google Scholar 
Velarde, E., et al., Nesting seabirds of the Gulf of California’s Offshore islands: Diversity, ecology and conservation. in Biodiversity, Ecosystems, and Conservation in Northern Mexico, Carton, J.-L. E., Ceballos, G., Felger, R. S. Eds. (Oxford University Press, 2005) pp. 452–470.Wilder, B. T., Felger, R. S. & Ezcurra, E. Controls of plant diversity and composition on a desert archipelago. PeerJ 7, e7286. https://doi.org/10.7717/peerj.7286 (2019).Article 
PubMed 
PubMed Central 

Google Scholar 
Ellis, J., Fariña, J. & Witman, J. Nutrient transfer from sea to land: the case of gulls and cormorants in the Gulf of Maine. J. Anim. Ecol. 75, 565–574 (2006).PubMed 
Article 

Google Scholar 
Wilder, B. T., Felger, R. S. & Morales, H. R. Succulent plant diversity of the Sonoran Islands, Gulf of California Mexico. Haseltonia 2008(14), 127–160 (2008).Article 

Google Scholar 
Lucassen, F. et al. The stable isotope composition of nitrogen and carbon and elemental contents in modern and fossil seabird guano from Northern Chile—Marine sources and diagenetic effects. PLoS ONE 12(6), e0179440. https://doi.org/10.1371/journal.pone.0179440 (2017).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Robinson, D. δ15N as an integrator of the nitrogen cycle. Trends Ecol. Evol. 16(3), 153–162 (2001).CAS 
PubMed 
Article 

Google Scholar 
Szpak, P., Longstaffe, F. J., Millaire, J.-F. & White, C. D. Stable isotope biogeochemistry of seabird guano fertilization: Results from growth chamber studies with maize (Zea mays). PLoS ONE 7(3), e33741. https://doi.org/10.1371/journal.pone.0033741 (2012).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Ezcurra, E., et al. Natural History and Evolution of the World’s Deserts. Global Deserts Outlook. United Nations Environment Programme (UNEP), 1–26 (2006).Yetman, D. The Great Cacti: Ethnobotany and biogeography (University of Arizona Press, 2007).
Google Scholar 
Álvarez-Borrego, S. Physical oceanography. in A New Island Biogeography of the Sea of Cortés, Case, T. J., Cody, M. L., Ezcurra, E. Eds. (Oxford University Press, 2002), pp. 41–59.Douglas, R., Gonzalez-Yajimovich, O., Ledesma-Vazquez, J. & Staines-Urias, F. Climate forcing, primary production and the distribution of Holocene biogenic sediments in the Gulf of California. Quatern. Sci. Rev. 26, 115–129 (2007).ADS 
Article 

Google Scholar 
Urbán, J. Marine mammals of the Gulf of California: An overview of diversity and conservation status. in The Gulf of California: Biodiversity and conservation, R. C. Brusca, Ed. (The University of Arizona Press and the Arizona-Sonora Desert Museum, 2010), pp. 188–209.Hastings, P. A., Findley, L. T., & Van der Heiden, A. M. Fishes of the Gulf of California. in: Brusca, R. C., (eds) The Gulf of California: Biodiversity and conservation 96–118, The University of Arizona Press and the Arizona-Sonora Desert Museum (2010).
Google Scholar 
Polis, G. A., Hurd, S. D., Jackson, C. T. & Sanchez Piñero, F. El Niño effects on the dynamics and control of an Island ecosystem in the Gulf of California. Ecology 78, 1884–1897 (1997).
Google Scholar 
Wilder, B. T. & Felger, R. S. Dwarf giants, guano, and isolation: The flora and vegetation of San Pedro Mártir Island, Gulf of California, Mexico. Proc. San Diego Soc. Nat. Hist. 42, 1–24 (2010).
Google Scholar 
Medel-Narvaez, A., Leon Luz, J. L., Freaner-Martinez, F. & Molina-Freaner, F. Patterns of abundance and population structure of Pachycereus pringlei (Cactaceae), a columnar cactus of the Sonoran Desert. Plant Ecol. 187, 1–14 (2006).Article 

Google Scholar 
Felger, R.S., Wilder, B.T. in collaboration with Romero-Morales, H. Plant Life of a Desert Archipelago: Flora of the Sonoran Islands in the Gulf of California. Tucson, University of Arizona Press (2012).Wilkinson, C. E., Hocking, M. D. & Reimchen, T. E. Uptake of salmon-derived nitrogen by mosses and liverworts in Coastal British Columbia. Oikos 108, 85–98 (2005).CAS 
Article 

Google Scholar 
Barrett, K., Wait, D. A. & Anderson, W. B. Small island biogeography in the Gulf of California: Lizards, the subsidized island biogeography hypothesis, and the small island effect. J. Biogeogr. 30, 1575–1581 (2003).Article 

Google Scholar 
Young, H. S., McCauley, D. J. & Dirzo, R. Differential responses to guano fertilization among tropical tree species with varying functional traits. Am. J. Bot. 98, 207–214 (2011).PubMed 
Article 

Google Scholar 
Nobel, P. S. Environmental Biology of Agaves and Cacti. Cambridge University Press (2003).Ramirez, K. S., Craine, J. M. & Fierer, N. Consistent effects of nitrogen amendments on soil microbial communities and processes across biomes. Glob. Change Biol. 18(6), 1918–1927 (2012).ADS 
Article 

Google Scholar 
Craine, J. M. et al. Ecological interpretations of nitrogen isotope ratios of terrestrial plants and soils. Plant Soil 396, 1–26 (2015).CAS 
Article 

Google Scholar 
Schoeninger, M. J. & DeNiro, M. J. Nitrogen and carbon isotope composition of bone collagen from marine and terrestrial animals. Geochim. Cosmochim. Acta 48(4), 625–639 (1984).ADS 
CAS 
Article 

Google Scholar 
Amundson, R. et al. Global patterns of the isotopic composition of soil and plant nitrogen. Global Biogeochem. Cycles 17(1), 1031. https://doi.org/10.1029/2002GB001903 (2003).ADS 
CAS 
Article 

Google Scholar 
Kahmen, A., Wanek, W. & Buchmann, N. Foliar δ15N values characterize soil N cycling and reflect nitrate or ammonium preference of plants along a temperate grassland gradient. Oecologia 156, 861–870 (2008).ADS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Bowen, T. Unknown Island: Seri Indians, Europeans, and San Esteban Island in the Gulf of California (University of New Mexico Press, 2000).
Google Scholar 
Evans, R. D. Physiological mechanisms influencing plant nitrogen isotope composition. Trends Plant Sci. 6(3), 121–126 (2001).CAS 
PubMed 
Article 

Google Scholar 
Dolby, G., Bennett, S. E. K., Lira-Noriega, A., Wilder, B. T. & Munguia-Vega, A. Assessing the geological and climatic forcing of biodiversity and evolution surrounding the Gulf of California. J. Southw. 57, 391–455 (2015).Article 

Google Scholar 
Case, T. J., Cody, M. L., & Ezcurra, E. A New Island Biogeography of the Sea of Cortés (Oxford University Press, 2002).Book 

Google Scholar 
Tershy, B. R. & Breese, D. The birds of San Pedro Mártir Island, Gulf of California Mexico. West. Birds 28, 96–107 (1997).
Google Scholar 
Tershy, B. R., Breese, D. & Croll, D. A. Human perturbations and conservation strategies for San Pedro Mártir Island, Islas de Golfo de California Reserve México. Environ. Conserv. 24, 261–270 (1997).Article 

Google Scholar 
Wilder, B. T. Historical biogeography of the Midriff Islands in the Gulf of California, Mexico. Dissertation. Riverside: UC, Riverside (2014).Post, D. M. et al. Getting to the fat of the matter: Models, methods and assumptions for dealing with lipids in stable isotope analyses. Oecologia 152, 179–189 (2007).ADS 
PubMed 
Article 

Google Scholar 
Kiljunen, M. et al. A revised model for lipid-normalizing δ13C values from aquatic organisms, with implications for isotope mixing models. J. Appl. Ecol. 43, 1213–1222 (2006).CAS 
Article 

Google Scholar 
Kuznetsova, A., Brockhoff, P. B. & Christensen, R. H. B. lmerTest Package: Tests in linear mixed effects models. J. Stat. Softw. 82(13), 1–26. https://doi.org/10.18637/jss.v082.i13 (2017).Article 

Google Scholar 
R Core Team, R: A language and environment for statistical computing. https://www.R-project.org/ (R Foundation for Statistical Computing, Vienna, Austria, 2022). More

Tsunekawa, A., Liu, G., Yamanaka, N. & Du, S. Restoration and Development of the Degraded Loess Plateau China 3–21 (Springer Press, 2017).
Google Scholar 
Kimura, R., Kamichika, M., Takayama, N., Matsuoka, N. & Zhang, X. C. Heat balance and soil moisture in the Loess Plateau, China. J. Agric Meteorol. 60(2), 103–113 (2004).
Google Scholar 
Deng, X. P., Lun, S., Zhang, H. P. & Turner, N. C. Improving agricultural water use efficiency in arid and semiarid areas of China. Agric. Water Manage. 80(1), 23–40 (2006).
Google Scholar 
Liu, C. A. et al. Maize yield and water balance is affected by nitrogen application in a film-mulching ridge-furrow system in a semiarid region of China. Eur. J. Agron. 52, 103–111 (2014).CAS 

Google Scholar 
Bu, L. D. et al. The effects of mulching on maize growth, yield and water use in a semi-arid region. Agric. Water Manage. 123, 71–78 (2013).
Google Scholar 
Hou, F. Y. et al. Effect of plastic mulching on the photosynthetic capacity, endogenous hormones and root yield of summer-sown sweet potato (Ipomoea batatas (L.) Lam.) in Northern China. Acta Physiol. Plant. 37, 164 (2015).
Google Scholar 
Jensen, K., Kimball, E. R. & Ricketson, C. L. Effect of perforated plastic row covers on residues of the herbicide DCPA in soil and broccoli. Environ Contam. Toxicol. B 35(6), 716–722 (1985).CAS 

Google Scholar 
Li, F. M., Guo, A. H. & Wei, H. Effects of clear plastic film mulch on yield of spring wheat. Field Crop. Res. 63(1), 79–86 (1999).
Google Scholar 
Liu, J. L. et al. Response of nitrogen use efficiency and soil nitrate dynamics to soil mulching in dryland maize (Zea mays L.) fields. Nutr. Cycl. Agroecosyst. 101(2), 271–283 (2015).CAS 

Google Scholar 
Li, R. et al. Effects on soil temperature, moisture, and maize yield of cultivation with ridge and furrow mulching in the rained area of the Loess Plateau, China. Agric. Water Manage. 116, 101–109 (2013).
Google Scholar 
Anzalone, A., Cirujeda, A., Aibar, J., Pardo, G. & Zaragoza, C. Effect of biodegradable mulch materials on weed control in processing tomatoes. Weed Technol. 24(3), 369–377 (2010).
Google Scholar 
Summers, C. G. & Stapleton, J. J. Use of UV reflective mulch to delay the colonization and reduce the severity of Bemisia argentifolii (Homoptera: Aleyrodidae) infestations in cucurbits. Crop Prot. 21(10), 921–928 (2002).
Google Scholar 
Chen, Y. S. et al. Empirical estimation of pollution load and contamination levels of phthalate esters in agricultural soils from plastic film mulching in China. Environ. Earth Sci. 70(1), 239–247 (2013).CAS 

Google Scholar 
Wang, S. Y. et al. Occurrence of macroplastic debris in the long-term plastic film-mulched agricultural soil: A case study of Northwest China. Sci. Total Environ. https://doi.org/10.1016/j.scitotenv.2022.154881 (2003).Article 
PubMed 

Google Scholar 
Hu, X. Y., Wen, B. & Shan, X. Q. Survey of phthalate pollution in arable soils in China. J. Environ Monit. 5(4), 649–653 (2003).CAS 
PubMed 

Google Scholar 
Zhou, X. Y. et al. Effects of residual mulch film on the growth and fruit quality of tomato (Lycopersicon esculentum Mill.). Water Air Soil. Pollut. 228(2), 1–18 (2017).ADS 

Google Scholar 
Hu, Q. et al. Effects of residual plastic-film mulch on field corn growth and productivity. Sci. Total Environ. 729, 1–10 (2020).
Google Scholar 
Wang, J. Z. et al. Crop yield and soil organic matter after long-term straw return to soil in China. Nutr. Cycl Agroecosyst. 102(3), 371–381 (2015).
Google Scholar 
Huang, Y. L., Chen, L. D., Fu, B. J., Huang, Z. L. & Gong, J. The wheat yields and water-use efficiency in the Loess Plateau: Straw mulch and irrigation effects. Agric. Water Manage. 72(3), 209–222 (2005).
Google Scholar 
Su, Z. Y. et al. Effects of conservation tillage practices on winter wheat water-use efficiency and crop yield on the Loess Plateau, China. Agric. Water Manage. 87(3), 307–314 (2007).
Google Scholar 
Ibrahim, A., Abaidoo, R. C., Fatondji, D. & Opoku, A. Integrated use of fertilizer micro-dosing and Acacia tumida mulching increases millet yield and water use efficiency in Sahelian semi-arid environment. Nutr. Cycl Agroecosys. 103(3), 375–388 (2015).CAS 

Google Scholar 
Zhang, D. K. et al. Suitable furrow mulching material for maize and sorghum production with ridge-furrow rainwater harvesting in semiarid regions of China. Agric. Water Manage. 228, 105928 (2020).
Google Scholar 
Myint, T. et al. Mulching improved soil water, root distribution and yield of maize in the Loess Plateau of Northwest China. Agric. Water Manage. 241, 106340 (2020).
Google Scholar 
Bai, Y. L. et al. Effects of long-term full straw return on yield and potassium response in wheat-maize rotation. J. Integr. Agric. 14(012), 2467–2476 (2015).CAS 

Google Scholar 
Liu, Z. J., Meng, Y., Cai, M. & Zhou, J. B. Coupled effects of mulching and nitrogen fertilization on crop yield, residual soil nitrate, and water use efficiency of summer maize in the Chinese Loess Plateau. Environ Sci. Pollut R. 24(33), 25849–25860 (2017).CAS 

Google Scholar 
Thomas, F. D., Michael, B., Jürgen, H., Maria, R. F. & Helmut, S. Effects of straw mulch on soil nitrate dynamics, weeds, and yield and soil erosion in organically grown potatoes. Field Crop. Res. 94(2–3), 238–249 (2005).
Google Scholar 
Tu, C., Ristaino, J. B. & Hu, S. J. Soil microbial biomass and activity in organic tomato farming systems: Effects of organic inputs and straw mulching. Soil Biol. Biochem. 38(2), 247–255 (2006).CAS 

Google Scholar 
Rao, Z. X. et al. Effect of rice straw mulching on migration and transportation of Cd, Cu, Zn, and Ni in surface runoff under simulated rainfall. J. Soils Sediment. 16(8), 2021–2029 (2016).CAS 

Google Scholar 
Ma, J., Xu, H., Yagi, K. & Cai, Z. C. Methane emission from paddy soils as affected by wheat straw returning mode. Plant Soil. 313, 167–174 (2008).CAS 

Google Scholar 
Xue, L. L. et al. Influence of straw mulch on yield, chlorophyll contents, lipid peroxidation and antioxidant enzymes activities of soybean under drought stress. J. Food Agric. Environ. 9(2), 699–704 (2011).
Google Scholar 
Wu, Y., Huang, F. Y., Jia, Z. K., Ren, X. R. & Cai, T. Response of soil water, temperature, and maize (Zea mays L.) production to different plastic film mulching patterns in semi-arid areas of Northwest China. Soil Tillage Res. 166, 113–121 (2017).
Google Scholar 
Blake, G. R. & Hartge, K. H. Bulk density. In Methods of Soil Analysis Part 1: Physical and Mineralogical Methods (ed. Klute, A.) 363–375 (American Society of Agronomy, Soil Science Society of America, 1986).
Google Scholar 
Li, F. M., Song, Q. H., Jjemba, P. & Shi, Y. Dynamics of soil microbial biomass and soil fertility in cropland mulched with plastic film in a semiarid agro-ecosystem. Soil Biol. Biochem. 36(11), 1893–1902 (2004).CAS 

Google Scholar 
Zhang, P. et al. Plastic-film mulching for enhanced water-use efficiency and economic returns from maize fields in semiarid China. Front. Plant Sci. 8, 512 (2017).PubMed 
PubMed Central 

Google Scholar 
Custodio, R. P. et al. The effects of increased temperature on crop growth and yield of soybean grown in a temperature gradient chamber. Field Crop. Res. 154, 74–81 (2013).
Google Scholar 
He, G. et al. Wheat yield affected by soil temperature and water under mulching in dryland. Agron. J. 109(6), 2998–3006 (2017).CAS 

Google Scholar 
Zhang, S. L. et al. Effects of mulching and catch cropping on soil temperature, soil moisture and wheat yield on the Loess Plateau of China. Soil Tillage Res. 102(1), 78–86 (2008).
Google Scholar 
Li, R., Hou, X. Q., Jia, Z. K. & Han, Q. F. Soil environment and maize productivity in semi-humid regions prone to drought of Weibei Highland are improved by ridge-and-furrow tillage with mulching. Soil Tillage Res. 196, 104476 (2020).
Google Scholar 
Wang, S. H. et al. Change in the bio-uptake of soil phthalates with increasing mulching years: Underlying mechanism and response to temperature rise. J. Clean Prod. 287(2021), 125049 (2020).
Google Scholar 
Li, W. W., Xiong, L., Wang, C. J., Liao, Y. C. & Wu, W. Optimized ridge–furrow with plastic film mulching system to use precipitation efficiently for winter wheat production in dry semi-humid areas. Agric. Water Manage. 218, 211–221 (2019).
Google Scholar 
Kader, M. A., Nakamura, K., Senge, M., Mojid, M. A. & Kawashima, S. Effects of colored plastic mulch on soil hydrothermal characteristics, growth and water productivity of rain-fed soybean. Irrig. Drain. 69(3), 483–494 (2020).
Google Scholar 
Luo, C. L. et al. Dual plastic film and straw mulching boosts wheat productivity and soil quality under the El Nino in semiarid Kenya. Sci. Total Environ. 738, 139808 (2020).ADS 
CAS 
PubMed 

Google Scholar 
Gouranga, K. & Ashwani, K. Effects of irrigation and straw mulch on water use and tuber yield of potato in Eastern India. Agric. Water Manage. 94(1), 109–116 (2007).
Google Scholar 
Lu, X. J., Li, Z. Z., Sun, Z. G. & Bu, Q. G. Straw mulching reduces maize yield, water, and nitrogen use in Northeastern China. Agron. J. 107(1), 406–414 (2015).
Google Scholar 
Zhou, L. F., Zhao, W. Z., He, J. Q., Flerchinger, G. N. & Feng, H. Simulating soil surface temperature under plastic film mulching during seedling emergence of spring maize with the RZ–SHAW and DNDC models. Soil Tillage Res. 197, 104517 (2020).
Google Scholar 
Chang, L. et al. Straw strip mulching affects soil moisture and temperature for potato yield in semiarid regions. Agron. J. 112(2), 1126–1139 (2020).CAS 

Google Scholar 
Zhang, P. et al. Effects of straw mulch on soil water and winter wheat production in dryland farming. Sci. Rep. 5(1), 209–222 (2015).
Google Scholar 
Ren, X. L., Zhang, P., Chen, X. L., Guo, J. J. & Jia, Z. K. Effect of different mulches under rainfall concentration system on corn production in the semi-arid areas of the Loess Plateau. Sci. Rep. 6(1), 47–50 (2016).
Google Scholar 
Akhtar, K. et al. Integrated use of straw mulch with nitrogen fertilizer improves soil functionality and soybean production. Environ. Int. 132, 105092 (2019).CAS 
PubMed 

Google Scholar 
Eden, G. R. S. M. The impact of organic amendments, mulching and tillage on plant nutrition, Pythium root rot, root-knot nematode and other pests and diseases of capsicum in a subtropical environment, and implications for the development of more sustainable vegetable farming. Australas. Plant Path. 37(2), 123–131 (2008).
Google Scholar 
Kader, M. A., Senge, M., Mojid, M. A., Takeo, O. & Kengo, I. Effects of plastic-hole mulching on effective rainfall and readily available soil moisture under soybean (Glycine max) cultivation. Paddy Water Environ. 15(3), 659–668 (2017).
Google Scholar 
Zhang, Z. et al. Plastic film cover during the fallow season preceding sowing increases yield and water use efficiency of rain-fed spring maize in a semi-arid climate. Agric. Water Manage. 212, 203–210 (2019).
Google Scholar 
Kader, M. A., Nakamura, K., Senge, M., Mojid, M. A. & Kawashima, S. Numerical simulation of water- and heat-flow regimes of mulched soil in rain-fed soybean field in central Japan. Soil Tillage Res. 191, 142–155 (2019).
Google Scholar 
Ryu, J. H. et al. Effects of straw mulching on soil physicochemical properties in Saemangeum reclaimed land. Korean J. Soil Sci. Fert. 49(1), 12–16 (2016).CAS 

Google Scholar 
Yin, W. et al. Growth trajectories of wheat–maize intercropping with straw and plastic management in arid conditions. Agron. J. 112(4), 2777–2790 (2020).
Google Scholar 
Wang, J. et al. Responses of runoff and soil erosion to planting pattern, row direction, and straw mulching on sloped farmland in the corn belt of northeast China. Agric. Water Manage. 25, 106935 (2021).
Google Scholar 
Cao, B. et al. Future landscape of renewable fuel resources: Current and future conservation and utilization of main biofuel crops in China. Sci. Total Environ. 806, 150946 (2022).ADS 
CAS 
PubMed 

Google Scholar 
Khawar, J. et al. Economic assessment of different mulches in conventional and water-saving rice production systems. Environ. Sci. Pollut. Res. 23, 9156–9163 (2016).
Google Scholar  More

Sites description and field operationsA total of six field studies were conducted in two different regions over two seasons. Four studies (two dryland and two irrigated) were in Kansas, United States (dryland: 39°4′30″ N, − 96°44′43″ W, irrigated: 39°4′25″N, − 96°43′12″ W) during the 2019 and 2020 growing seasons (hereafter referred to as USDry19, USIrr19, USDry20, and USIrr20 studies). The remaining two studies (dryland) were in Entre Rios, Argentina (31°50′49″ S; 60°32′16″ W) during the 2018/2019 and 2019/2020 growing seasons (hereafter referred to as Arg19 and Arg20 studies). The soils were Fluventic Hapludolls [silt loam, 40% sand, 13% clay, 47% silt, organic matter (OM) 1.7%, 7.7 pH, 31.1 ppm P (Bray−1)] at the US dryland studies, and Pachic Argiudolls [silty clay loam, 10.1% sand, 30.6% clay and 59.3% silt, OM 3.2%, 6.8 pH, 34.7 ppm P (Bray−1)] at the US irrigated studies. At the Argentinian studies soil was a Vertic Argiudoll in 2019 [silty clay loam to clay loam, 3.9% sand, 27.6% clay, 67.9% silt, OM 2.65%, 7.2 pH, 12.5 ppm P (Bray−1)] and an Acuic Argiudoll in 2020 [silt loam to silty-clay-loam, 5.6% sand, 28.6% clay, 65.8% silt, OM 3.33%].The US dryland and irrigated studies were sown on June 4, 2019, and May 20, 2020. In 2019, the dryland study was replanted on June 29 due to poor emergence after the first sowing. The studies in Argentina were sown on December 5 in 2018 and November 20 in 2019. At all six studies, plots were kept free of weeds, pests, and diseases through recommended chemical control.The genotypes used in the US were P40A47X (MG 4.0) and P39A58X (MG 3.9) (Corteva Agriscience, Johnston, IA, USA) in 2019 and 2020, respectively. Both varieties are tolerant to glyphosate and dicamba herbicides (RR2X) and have low lodging probability. For the northeast region of Kansas, recommended sowing dates range from May 15 to June 15 along with MG 421. In addition, recommended seeding rates are between 270 and 355 thousand seeds ha−1 for low-yielding environments and 190 to 285 thousand seeds ha−1 for medium- and high-yielding environments13. In Argentina, the genotype AW5815IPRO (MG 5.8, Bayer, Leverkusen, Germany) was used both in 2020 and 2021, it is tolerant to glyphosate and sulfonylureas, and has low lodging probability. Recommended sowing dates for Entre Rios considering soybeans as a single crop range from October 20 to December 10, and MG usually range from 4 to 6; lastly, seeding rate recommendations are between 200 and 250 thousand seeds ha−1 in the region22.Study designThe studies carried out in the US were arranged as a split plot design with three replicates in both 2019 and 2020. In 2019, the main plot treatment factor was planter type with two levels [John Deere (Moline, Illinois, US) Max Emerge planter (ME, 12 rows), and John Deere Exact Emerge Planter (EE, 16 rows)], and the split-plot treatment factor was seeding rate with two levels (160 and 321 thousand seeds ha−1). In 2020 the main plot treatment factor was also planter type with two levels (ME and EE), and the split-plot treatment factor was seeding rate with four levels (160, 215, 270 and 321 thousand seeds ha−1). Planting speed was 7 km h−1 in both studies and years, plots were 24 and 32 rows wide when planted with ME and EE, respectively, with 0.76 m row spacing. Plot length was 80 m in the dryland studies and 160 m in the irrigated studies. The studies in Argentina were arranged as a single replicate of each seeding rate (100, 230, 360 and 550 thousand seeds ha−1) in both years. Planting speed was 5.5 km h−1 in both years, and plots were 10 rows wide with 0.52 m row spacing and 350 m in length.All treatment factors in US studies were evaluated with the overall goal of producing substantial variation in the variable of interest, plant-to-plant spatial uniformity, rather than to make an inference of their effect on yield. The Argentinian studies were only used for selection of stand uniformity variables due to the single replicate. Plant spatial uniformity variables were first fitted using the data from US studies (details below), and then the best explanatory metrics were selected to re-fit the relationships combining both data sets from US and Argentina. Finally, sowing dates, maturity groups, and seeding rates evaluated in this study at both locations (Arg and US) were aligned with those recommended for each region.Data collection and spacing uniformity variablesTwo segments of 2 m in length were established early in the season inside each plot. At the V5 (US studies) and R1 (Arg studies) soybean development stage23, the cumulative distance of the plants within each segment was measured and then used to calculate multiple derived variables. Plant spacing (cm) was calculated as the average distance between neighboring plants. In addition, the distance from a plant to each neighboring plant was classified as shorter or longer than the plant spacing (named nearest and farthest neighbor distance, respectively). Achieved versus Target Evenness Index (ATEI, dimensionless) was calculated as the ratio between the observed plant spacing and the theoretical plant spacing (TPS, cm), where TPS is the expected plant spacing derived from a specific seeding rate and row width (Eq. 1).$$ATEI = frac{Spacing;(cm) }{{TPS;(cm)}}$$
(1)
The ATEI index was designed to account for the proximity of the observed plant spacing to the TPS. Values closer to 1 indicate that the plant spacing is close to the TPS and values that are below or above 1 indicate that the plant spacing is lower or higher than the TPS, respectively; thereby departing from an ideal plant spacing. Hence, ATEI values greater than 1 depict both (i) non-uniform plant-to-plant spacing distribution and (ii) plant densities below the target (seeding rate). To further understand the meaning of ATEI, the relative density (rd) was calculated as the ratio between plant density (based on the number of plants in the 2 m segment) and seeding rate.To account for the unevenness of distance from a plant to both neighboring plants within the row, we used the Evenness Index (EI, dimensionless), calculated as the ratio between the distance to the nearest neighbor (cm) and the plant spacing (cm) of a given plant (Eq. 2). The Evenness Index values range from 0 to 1, a value closer to 1 indicates that a plant is equidistantly spaced to both of its neighboring plants within the row, if zero then those plants are occupying the same position (as doubles). It is important to note that EI does not provide information on the spacing (in distance, cm) or how close the spacing is compared to the TPS, but only describes the unevenness distance of a plant to its neighboring plants within a row.$$Evenness ;Index; (EI) = frac{nearest; neighbor ;(cm)}{{Spacing; (cm)}}$$
(2)
In addition, the distance from a plant to its preceding neighboring plant, and the TPS were used to classify the position of each plant into one of eight classes (Fig. 1). Plants were classified in classes ranging from “double” (preceding plant distance  Double-skip) as a function of seeding rate, planter type and their interaction (fixed effects), and block nested in site-year (random effect) (Tables 1 and 2). Independent models for each of the 4 US studies were built assessing the effects of planter type, seeding rate, and their interaction (fixed effects), and seeding rate nested in planter type, and in block (random effects) on the same variables previously mentioned (Supplementary Table 1). The models were run using the lmer function from lme4 package in R (R Core Team, 2021). In addition, the US and Arg studies were combined to evaluate the effect of site-year on yield, plant density, and all stand uniformity variables (Supplementary Fig. 1) using the lm function from package stats. Means separation were performed using Fisher’s LSD (Least Significance Difference) test (alpha = 0.05) with emmeans function from package emmeans.Table 1 Effect of planter type, seeding rate, and their interaction on variables from plant position classification for all US studies. References: percentage of perfectly spaced plants (Perfect), percentage of plants misplaced by 66% (Mis 66), percentage of plants misplaced by 33% (Mis 33), percentage of double plants (Double), percentage of short skips plants (Short-skip), percentage of long skip plants (Long-skip), percentage of double skips plants (Double-skip), and percentage of greater than double skip plants ( > Double-skip).Full size tableTable 2 Effect of planter type, seeding rate, and their interaction on yield and stand uniformity variables for all US studies. References: Spacing between plants standard deviation (Spacing sd), achieved versus targeted evenness index mean and standard deviation (ATEI and ATEI sd, respectively), and evenness index mean and standard deviation (EI and EI sd, respectively).Full size tableCommunity-scale data from the four US studies were combined and fitted to bivariate linear regression models with yield as the response variable and each of the stand spatial uniformity variables as the explanatory variable. Significant models (alpha = 0.05) were further evaluated by calculating the coefficient of determination (R2) and root mean squared error (RMSE) (Fig. 2). Models with the lower RMSE and higher R2 were selected as those that best captured the effect of non-uniform stands on soybean yield. After variables were selected, both US and Arg data sets were combined and the linear regressions between the selected variables and yield were re-fitted to assess the consistency of the relationships when an independent data set was included. Community-scale yield from US and Arg studies was modelled as a function of the selected stand uniformity variable, country (US and Arg), and their interaction (fixed effects) (Fig. 3). The spatial uniformity metric showing the most consistent relationship for both US and Arg studies (i.e., non-significant interaction between stand uniformity metric and country), was selected to continue the analysis. The bivariate linear regression models were run with function lm.Figure 2Relationship between stand uniformity variables and soybean yield for US studies. ATEI mean and sd achieved versus targeted evenness index mean and standard deviation, EI mean and sd evenness index mean and standard deviation, Perfect percentage of perfectly spaced plants, R2 coefficient of determination, RMSE root mean square error. All stand uniformity variables presented a significant slope at alpha = 0.05.Full size imageFigure 3Relationship of spacing standard deviation (Spacing sd, cm) and achieved versus targeted evenness index standard deviation (ATEI sd) to soybean yield. Different colors and line types denote different countries (Argentina, Arg—full line, red points; United States, US—dashed line, blue points). R2 coefficient of determination, RMSE root mean square error.Full size imageDifferent environmental conditions and seeding rate levels may modify the effect of plant spatial uniformity on yield. To explore this, each of the studies from Arg and US were separated into low- (USDry19 and ArgDry20, mean of 2.7 Mg ha−1), medium- (USIrr19, USDry20 and ArgDry19, mean of 3.0 Mg ha−1), and high- (USIrr20, mean of 4.3 Mg ha−1) yield environments based on the effect of site-year on yield (Supplementary Fig. 1). Additionally, the tested seeding rates were separated in low ( 300 thousand seeds ha−1) levels based on the current optimal seeding rate for medium yielding environments (235 thousand seeds ha−1, 4 Mg ha−1)13 and the extreme values proposed by Suhre et al.11 (148 and 445 thousand seeds ha−1). This classification was used to model yield as a function of (i) the selected stand uniformity metric, yield environment, and their interaction, and (ii) the selected stand uniformity metric, seeding rate levels, and their interaction. These models were tested to obtain a robust conclusion on the overall effect of yield environment and seeding rate levels, and their interactions (all treated as fixed effects) with plant-to-plant spatial uniformity relative to the response variable, soybean yield. The Akaike information criteria (AIC) was used to compare the full (with interactions) relative to the reduced models (single effects).Ethics declarationsExperimental research and field studies on plants including the collection of plant material, complied with relevant institutional, national, and international guidelines and legislation. More