Ecology

Simulation experimental set upLaboratory monitoring of leakage migration process can provide an important basis for field tests. The designed and improved ERT device can better describe the migration range of leakage in soil41. In this experiment, a parallel potential monitoring device was used to improve the monitoring of leakage fluid migration. The simulation experiment in the laboratory is carried out in a (100 cm*100 cm*50 cm) plexiglass tank. Sand and clay shall be screened with a 2.36 mm square sieve, watered and compacted with a board to ensure that the soil layer is in close contact with the measuring electrode.Electrode arrangementThe ground wire of high-density electrical method instrument is connected to the electrodes arranged around the bottom of the tank as the power electrode C2, as shown in Fig. 2a. The host is connected to the electrode system. The electrode system consists of 47 electrode grids with a spacing of 0.08 m. The measuring electrode P1 is connected to the mainframe through a wire 0.05 m below the grid center. The geomembrane is located 0.03 m above the measuring electrode P1. The collection device is used as a monitoring system for various leachate. The arrangement of electrodes is shown in Fig. 2b. The power supply electrode C1 is placed at a certain depth in the middle of the saturated sand to provide a constant current. The location of electrode C1 and leakage point is shown in Fig. 2c. The layers from the bottom of the tank are silty clay, geomembrane, silty clay and saturated sand, as shown in Fig. 2d.Figure 2Set-up of leachate migration simulation experiment: (a) Schematic diagram of electrode C2 layout; (b) Schematic diagram of electrical system laying; (c) Position of electrode C1 and leakage point; (d) Schematic diagram of simulated experimental soil layer.Full size imageComposition of monitoring systemThe electrode system is used to monitor the background electric field and artificial electric field of the landfill site. In the experiment, the electrode system is laid in the clay layer under the geomembrane. It is composed of detection electrodes distributed in a grid at a certain distance.The electrical signal conversion system adjusts the measurement mode, sampling accuracy, acquisition frequency and other parameters of the electrode in the field according to the instructions of the mainframe, and transmits the collected electrical signal to the mainframe.The mainframe can control the operation of the monitoring system. The possible leachate points and their pollution range are determined by collecting data. The system mainly includes mainframe and its software system, power supply, etc., as shown in Fig. 3.Figure 3Se2432 parallel electric method instrument.Full size imageLeachate devicePlace 4 leakage bottles above the tank. No.1 and No.4 bottled water are used to simulate the leakage liquid formed by the direct infiltration of rainwater in slag through geomembrane and as a reference. Because Cl-1 is a typical pollutant in the landfill. No. 2 bottle containing 20 g/L NaCl solution is used to simulate inorganic salt leakage in urban life. No. 3 bottle containing 20 ml/L ethanol solution is used to simulate the leakage liquid containing a large amount of organic matter in municipal solid waste. The characteristics of leachate have been summarized in Table1.Table 1 The characteristics of leachate.Full size tableBefore the experiment, configure four solutions, close the injection, use an electric meter to check the conductivity of each measuring point. After each measuring point has no open circuit, supply power to the soil layer through the mainframe to measure the background electric field of the soil. Then open the injection, adjust the flow rate, release the solution at a fixed flow rate, record the soil electric field in the process of leakage every half an hour, collect the potential values of each measuring point, process the data through the potentiometry and potential difference method, and form the relevant potential horizontal profile and longitudinal section of the soil.Principle of potentiometric detection technologyWhen there are leakage points in the landfill, power is supplied to the landfill, and the current forms a current loop through the geomembrane. If there are n (n = 1,2,3…) leakage points in the geomembrane, the power supply current is I, and the artificial electric field will form a leakage electric field at the leakage point, which can be used as a point power supply.$$I = int dI = int j cdot dS$$
(1)
where I is the current intensity, j is the current density vector, and S is the area passing through the current.When there are n leakage points, I will be shunted. If a leakage point is regarded as a finite surface, the current intensity I as:$$I = {I_1} + {I_2} + cdot cdot cdot + {I_{text{n}}} = sumlimits_{i = 1}^n {int_{S_i} {jdS} }$$
(2)
Generally, the power supply current field of landfill site will be affected by the formation medium structure. It is assumed that the formation medium structure is composed of three layers, each layer has uniform properties and stable conductivity, and the layers from top to bottom are: landfill layer, with resistivity of ρ1. The saturated leakage liquid layer above the geomembrane has a resistivity of ρ2. The clay layer under the geomembrane has a resistivity of ρ3. The electrode C1 is arranged in the garbage layer for power supply, and the electrode C2 is arranged at the lower part of the geomembrane away from the electrode system area. The electrode C2 can be regarded as a far pole.Because of the ρ1  > ρ2, the conductivity of the saturated leakage liquid layer at the upper part of the geomembrane is better than that of the landfill layer, so that there is almost no reflected current between the ρ1 layer and the ρ2 layer, that is, the current generated by the power supply electrode C1 is all transmitted to the ρ2 layer. Because of the ρ3  > ρ2, it can be considered that the interface between ρ2 layer and ρ3 layer has both a reflection current, and a transmission current through the leakage point. The potential generated at the detection electrode P1 under the geomembrane is formed by the action of transmission current. The total potential of point P1 is obtained by the superposition of the potential of point power supply passing through n leakage points at P1.$${U_{P1}} = sumlimits_{i = 1}^n {frac{{{I_i}{rho_3}}}{{2pi {{text{r}}_{iP1}}}}}$$
(3)
Parallel potential difference methodThe test adopts pole–pole arrangement, and the calculation formula of apparent resistivity is as follows:$$rho = 2pi {text{aR}}$$
(4)
where ρ is apparent resistivity; a is the distance between electrodes C1 and P1; R is measuring resistivity.When there are loopholes in the geomembrane of the landfill, the leakage liquid will gradually penetrate into the soil layer under the geomembrane through the loopholes, resulting in the change of the conductivity of the soil layer under the geomembrane. The pole-pole acquisition mode of Se2432 parallel electrical instrument is used to obtain the original data (potential difference) of each measuring point on the grid. After current normalization, the apparent resistivity of the soil layer is obtained. The electrical properties of different depths of the soil layer can be obtained by inversion of the apparent resistivity data of the soil layer, so as to determine the occurrence point and distribution range of leakage.The monitoring grid is 5 × 5. The spacing between measuring points is 0.08 m. The measurement method adopted by Se2432 parallel electric method instrument is cross diagonal measurement method. Figure 4 shows that it only needs to measure the potential values on the measuring points on the horizontal, vertical and 45° diagonal lines.Figure 4Schematic diagram of cross-diagonal measurement method.Full size imageTheoretical calculation of test modelTheoretical results of 10 × 10 grid monitoringAccording to the experimental model and statistical data, the resistivity of the clay layer under the geomembrane is assumed ρ = 10 Ω· m, the resistivity ratio of tap water, NaCl solution and ethanol solution after penetrating into the soil layer ρNo.1:ρNo.2:ρNo.3 = 5:3:10. If the four leakage points set by the model are regarded as four conductive resistors, the ratio of the current passing through the four leakage points is INo. 1:INo. 2:INo. 3:INo. 4 = 6:10:3:6.The calculation model is 10 × 10 grid, and the spacing of measuring points is 0.05 m. The potential value on each measuring point is calculated according to Eq. 3, and the obtained data is processed with surfer software to obtain the potential contour map, as shown in Fig. 5. Among them, points 1, 2, 3 and 4 are the leakage positions of water, NaCl solution, ethanol solution and water respectively, and the spacing between leakage points is 0.15 m.Figure 510 × 10 Grid theory detection potential contour map.Full size imageFigure 5 shows that the leakage fields formed by the four kinds of leaking liquids interfere with each other from the theoretical calculation results. The leachate current at point 2 is larger, the high potential closed loop is obvious, and its center corresponds to the leakage center. The reason for this is that the NaCl solution contains conductive particles that increase the conductivity of the leak point. Point 1 and 4 are the same as water, and the leakage electric field is almost the same. Its closed loop is obvious, and the high potential center also corresponds to their leakage position. There is almost no closed loop effect at point 3 under the influence of 1, 2 and 4. The results show that the leakage field formed by high resistance leakage liquid is not easy to be detected by potentiometric detection, and low resistance leakage is suitable to be detected by potentiometric detection.Theoretical results of 12 × 12 grid monitoringThe resistivity of the clay layer under the geomembrane is assumed ρ = 10Ω·m. In consideration of the mutual influence between the leachate and appropriately reduce its influence effect, the resistivity ratio of water, NaCl solution, and ethanol solution after penetrating into the soil layer is set as ρNo.1:ρNo.2:ρNo.3 = 20:15:24, the ratio of the current passing through the four leakage points is INo.1:INo.2:INo.3:INo.4 = 6:8:5:6. And adjust the distance between the two points to 0.28 m. 12 × 12 grid was used for detection, and the spacing of detection points is 0.04 m. Calculate the potential value of each detection point according to Eq. 3, and use Surfer to obtain the detection contour map of four kinds of leakage, as shown in Fig. 6.Figure 612 × 12 Grid theory detection potential contour map.Full size imageTheoretical calculation results show that when the distance between the leakage points is large and the distance between the detection points is small, the potentiometric method can detect the leakage position of various leachates well. At the same time, the diffusion range of different leachates in the same plane is roughly the same, and they all gradually diffuse outward from the center of the leakage point, and the potential value gradually decrease. Point 2 has the largest potential closed loop range, which also has a certain impact on the leakage points of adjacent points 1 and 3. Point 1 and point 4 are water leakage. Affected by different leakage liquids, the leakage electric field of the two same leakage liquids is obviously different. The potential closed loop range of point 1 is larger than that of point 4. Point 3 is the leakage of ethanol solution. Because its resistance is the largest, the range of potential closed loop is the smallest.Figure 7 shows that the leakage fields around the leachates are funnel-shaped, and its size is related to the type of leachate. Therefore, different network density should be designed for different types of leakage liquid, so as to use the most economical scheme to detect the leakage point.Figure 712 × 12 Grid theory detects potential 3d view.Full size imageInterpretation and discussion of resultsLaboratory simulation experiment researchFigure 8a shows the background electric field potential of soil layer. The four injection pipes are opened at the same time and adjusted to the same flow rate. Under the condition of continuous leakage, monitor the leakage field potential at an interval of 1 h. Figure 8b shows the leakage electric field potential value for 1 h. Reduce the injection pipes flow rate to 1/2 of the initial value. Figure 8c shows the monitoring results of 2 h soil layer leakage field potential. Figure 8d shows the soil leakage field potential monitored after 30 min of sealing the injection pipes.Figure 8Leakage field potential diagram of soil layer: (a) Background electric field of soil layer; (b) Potential distribution of soil layer after 1 h of leakage; (c) Potential distribution of soil layer after 2 h of leakage; (d) Potential distribution of soil layer after closing the injection tube for 30 min.Full size imageFigure 8a shows that the background potential contour of the experimental soil layer is at a lower value. Few current lines pass through the monitoring area. A dense closed potential circle of high potential value is formed at point 2. The current flow at point 2 is greater than the other points 1, 3 and 4. The analysis result may be that in the process of watering and compaction, the clay layer under the geomembrane is not uniform, and the compaction degree of the soil layer is different, resulting in different potential values ​​obtained by monitoring. The permeability at point 2 is better than other points, so when the flow rate of the leakage liquid is large, the leakage liquid under the geomembrane gathers near point 2 and spreads out around. After the clay is watered and compacted, the soil compaction is small and the pore water content is large, resulting in a high potential abnormal area in the lower left corner of point 3.Point 2 forms a closed loop of anomaly potential contour much higher than the background electric field, while the value of potential contour coil at leakage point 3 is lower than the surrounding value. It can be analyzed that positions 2 and 3 are leakage points. The leachate at point 2 is a high concentration NaCl solution containing more conductive particles, which will reduce the resistivity of the soil layer under the geomembrane at point 2. Thus, the passing current is increased to form a high potential closed loop. The leachate at point 3 is ethanol solution, which will increase the resistivity of the soil layer under the geomembrane at point 3. So as to reduce the passing current and form a low potential closed loop. Figure 8b shows that the potential contour is consistent with the influence of NaCl solution and ethanol solution on the soil layer under the geomembrane. It can be concluded that point 2 and point 3 are leakage points. The electric field formed after water leakage at point 1 and point 4 cannot clearly distinguish the leakage points.During the monitoring process, the leachate was continuously released from the injection pipe, and the results reflected the dynamic characteristics. Figure 8b shows the phenomenon that the leachate from point 1 and point 4 aggregates around point 2, which is consistent with the inference of better permeability at point 2. Figure 8b,c show that when the flow rate of the leachate is changed and the flow rate of the injection pipe is reduced, the high-potential region of the entire electric field is reduced. Under the influence of gravity, the leachate will migrate longitudinally, and the closed-loop abnormally high-potential regions and abnormally low-potential regions at points 2 and 3 also decrease.Compared with the surrounding potential contours, the difference is more obvious. Figure 8d shows that when the injection pipe stops leaking for a period of time, the leachate migrates longitudinally along the leakage point. At this time, the electric field of the soil layer is similar to the original background electric field, but the potential value is higher than the background electric field, indicating that the leachate is stagnant in the pores of the soil layer, it is the result of changing the electrical properties of the soil layer. The parallel potential method can collect the potential value of each point in the field at one time, which provides a basis for real-time monitoring of landfill leachate.Figure 9 shows the inversion results of the horizontal section of the experimental model. The blue area corresponds to the distribution range of the low resistance anomaly. There are no jump or distortion points in the profile. The resistivity in the longitudinal direction basically shows a change from low to high. The upper layer seepage liquid migrates, and the bottom soil layer is characterized by low humidity and high resistivity. The low-resistance areas formed by the leakage of NaCl solution are widely distributed in the horizontal section. The distribution range is 0–0.28 m, and the migration scale of the leakage liquid can be clearly seen. The morphological characteristics of water leakage in different parts are basically the same. The distribution range is 0–0.18 m. The leakage of ethanol solution is only reflected at 0–0.06 m, and the distribution range is the smallest at the same depth. The ethanol solution also had the slowest migration rate.Figure 9Inversion map of plane section at different depths.Full size imageFigure 10 shows the inversion results of the X–Z longitudinal section of the test model. The two apparent resistivity profiles at Y = 0.24 m and Y = 0.32 m show that there is no low-resistance area in the shallow layer on the soil layer, indicating that the geomembrane in this area is not damaged. The low resistance zone in the middle is caused by the lateral migration of leakage fluid. The low-resistance anomaly area at the top of the profile can be judged as a leak point or formed by the migration of nearby leachate. Combined with the horizontal section, the leakage depth is similar, and the lateral migration speed of leachate is faster than the longitudinal migration speed. Four leak points can be distinguished, delineating the general location of the leak.Figure 10X–Z longitudinal section on different Y axes.Full size imagePhysical model experimentThe potential value of each electrode was monitored after 2 h of leakage, and the resistivity profiles at different positions were obtained by the potential difference method.It can be seen from Fig. 11 that the potential difference method can monitor the leakage of leachate in different directions. The morphological features of the plume formed by the downward migration of the leak point are approximately funnel-shaped in longitudinal section. The affected area of ​​the soil layer can be obtained in time. Figure 11b shows that the potential difference at the monitoring point is very different on both sides. After 2 h of leakage, a large amount of leakage liquid exists in the soil layer. When the water content in the soil layer increases, the diffusion rate of the ethanol solution increases, showing high resistance characteristics. At the same time, due to the action of gravity, there is a lot of vertical migration, and the potential value changes greatly. The profile clearly shows that the distribution area of ​​high potential difference is large, and the distribution of low potential is small. Figure 11c shows that since the migration rate of leachate in the horizontal direction is greater than that in the vertical direction, the potential difference of the monitoring point in the middle region is smaller, and a closed region of a high-potential circle is formed in the middle. The difference between the two results in a smaller potential difference area. Figure 11d shows that almost all the low-potential areas on the monitoring point are on the left side, because the leakage rate of NaCl solution in the horizontal direction is similar to that in the vertical direction under the condition of good soil compaction. At this time, a large number of conductive particles are contained, resulting in a large high-potential region. The difference between the two forms a large area of ​​low potential difference on the left. This is in good agreement with the lower resistance characteristics of the NaCl solution. Figure 11e shows that the two low-resistance regions correspond to the two leakage centers. The low potential difference region is formed by migration around the leak point. The migration speed in the horizontal direction is similar to that in the vertical direction, and the water migration speed on the left is slower than that of the sodium chloride solution on the right. Figure 11e,f show that the monitoring results are the same, but the resulting potential difference is also increased. This is affected by the distance between the monitoring point and the leak point. When the monitored point and the leakage point are located on the same section, the soil layer is the most severely affected area by leakage. Through the change of the potential difference, the leakage range and the location of the leakage point can be better judged.Figure 11Electrical resistivity tomograms of profile: (a) Resistivity of the slitting profile Y = 0; (b) Resistivity of the slitting profile Y = 0.08; (c) Resistivity of the slitting profile Y = 0.16; (d) Resistivity of the slitting profile Y = 0.24; (e) Resistivity of the slitting profile Y = 0.32; (f) Resistivity of the slitting profile Y = 0.4.Full size image More

Phakamani M’Afrika Xaba speaks at a botanical workshop.Credit: Nong Nooch/Tropical Botanical Garden

For Black communities in today’s South Africa, the legacies of colonialism and apartheid still prevail, shaping social structure and limiting access to opportunities. Colonialism displaced Black South Africans from the mid-seventeenth century, eroding cultural and social systems.From the 1950s, apartheid legalized systematic racial discrimination against disenfranchised, mainly Black, people. It limited their economic opportunities and social standing, prescribing an inferior education system to deliberately shape a poor quality of life. The policy fuelled systemic sexism, sexual-orientation discrimination, ageism, and the use of ethnicity as a divide-and-conquer strategy.Seventy years later, even after more than 25 years of democracy following the end of apartheid in 1994, schools and suburbs are still predominantly segregated, with government funding unevenly allocated in terms of facilities and quality of education.Former South African president Nelson Mandela once said, “In Africa there is a concept known as ubuntu — the profound sense that we are human only through the humanity of others; that if we are to accomplish anything in this world, it will in equal measure be due to the work and achievement of others.”As three past and present employees of the South African National Biodiversity Institute (SANBI), a conservation organization founded in 2004 to manage the country’s biodiversity resources, we have been advocating for a culture of treating others in the way we want to be treated: by applying universal shared human values, redefining institutional culture and systems to be inclusive, and opening safe spaces for a diversity of ideas. We have proposed a ground-up approach that aims to focus on holistic transformation at different levels in our organization.Our approach was to initiate a platform to identify inclusivity challenges, foster awareness and collaboration among staff and collectively develop innovative ideas and solutions. These would be aligned to existing organizational values, such as ubuntu, growth, respect and tolerance, excellence, accountability and togetherness. We strive to bring about institutional cultural change through facilitated, constructive conversations, by strengthening connections and cohesion among staff and through creative and proactive problem-solving across our institution.Mentorship that thrivesInstitutional culture needs to enable successful mentoring by creating a safe space. For example, SANBI’s mentoring programme for interns, students and early-career scientists involves quarterly meetings between them and dedicated human-resources staff — check-ins that provide a space to engage with programme coordinators without early-career researchers’ supervisors being present. In addition to sharing feedback on institutional policies and procedures, early-career scientists have the opportunity to discuss challenges they might face because of their supervisor or work placement. When issues are identified early, transfers or exchanges between work programmes can be arranged.Every year, we each sign up to mentor junior researchers to provide a supportive environment for guidance, counselling and the transfer of skills. We develop structured workplans with specific goals and outputs, and we discuss expectations with our protégés. In addition, we offer shared workspaces for interns and encourage peer learning, so that interns can form a peer support network. In these relationships, trust is crucial and can be a gateway to broader professional and personal networks.

Early-career researchers doing fieldwork training at the Stellenbosch University Experimental Farms in South Africa.Credit: Tlou Masehela

Institutions should recruit outside of their walls, if necessary, to ensure that appropriately skilled mentors are paired with early-career researchers. For mentorship to thrive, institutions must also create an enabling environment. In non-supportive environments, staff — particularly those from under-represented groups — who remain inadequately skilled and work without guidance become frustrated. Some can even feel they don’t belong because they see themselves as lagging behind their peers.Institutions often focus too strongly on outputs — such as publications, products or technologies — at the expense of reflecting on the values that uphold the institution. These values might be outdated and out of touch with those of staff, or with those held by partners, stakeholders or society at large. If staff cannot relate to the institutional culture and systems, job satisfaction and retention rates can suffer.Until a few years ago, for example, venues at our organization were named after former staff, as a way of acknowledging their contributions. Inevitably, these were mostly white, male, senior staff, such as Harold Pearson, the first director of Kirstenbosch National Botanical Garden, and Brian Rycroft, who served as director in the 1950s. But the contributions of staff who were employed in junior positions for 20–30 years also needed to be acknowledged. After an outcry around 2014, then-chief-executive Tanya Abrahamse, the first Black woman to hold the post, decided to acknowledge contributions of staff no matter their position. As a result, we now have Richard Crowie Hall, an exhibition space named after one of SANBI’s longest-serving staff members.The protracted legacy of apartheid in South Africa means that if institutional implicit biases are left unaddressed, they can create a fertile ground for racial, ethnic, tribal, financial and gender tensions. We urge more institutional recognition of the contributions of all.Fostering safe spacesThrough our engagements with each other, we have discovered a set of shared values, aligned with those of our institution, and have set out to establish a space to build our vision of a supportive, safe environment based on these values. Safe spaces are required for expressing controversial or uncomfortable views and to do the hard work of finding solutions to inequities. Confidentiality and trust cultivate such safe spaces, which can be created initially in small groups, then expanded to constructive formal or informal spaces. The conversations and suggestions of informal discussion groups about staff development and transformation can be elevated to management for implementation.
Decolonizing science toolkit
Safe spaces are a necessity for institutions that wish to truly address their legacies of racism and colonialism. Policies alone will not create these spaces — they require dedicated staff, too. Such spaces should ensure that those who speak up can do so without fear of being labelled as troublemakers or victimized.As a first step in pursuing this vision, we met with the senior teams at our organization to share ideas around the need for and benefits of an inclusive culture. We highlighted that inclusivity improves work–life balance, productivity and mental well-being for all employees.Any change, transformative or otherwise, is a process that takes perseverance, patience and determination. For any individual scientist to grow and flourish, they need a supportive environment, rich mentorship, a safe space and an enabling culture. It’s time for those factors to apply to all scientists equitably, no matter their gender, race, ethnicity or tribe. By fostering this mindset, we aim to reframe the narrative of our history and, in doing so, give all South African scientists their chance to thrive. More

Variation in environmental parameters across space and time in Bohai BayThe environmental parameters of samples collected near the Tianjin coastal area from different stations and seasons exhibited high temporal and spatial heterogeneity. The seawater temperature was 28.09 ± 0.53 °C in Aug, 17.48 ± 2.36 °C in May, and 19.55 ± 1.26 °C in Oct (Table 1). The seasonal variation in seawater temperature corresponded to the meteorological characteristics in Bohai Bay, with warm seawater in summer and relatively cool seawater in spring. The salinity was 29.69 ± 2.71‰ in Aug, 33.19 ± 0.33‰ in May, and 30.15 ± 1.63‰ in Oct. Seasonal variations in salinity may be mainly related to freshwater loading. According to the precipitation observed data of Bohai Bay in previous years, the rainfall amount and days in summer are the most19, which may lead to the increase in runoff and the relatively low salinity in summer. Chlorophyll a (Chl a) was highest in May, with lower levels in Aug and Oct. The dissolved inorganic nitrogen (DIN) was significantly higher in May and Aug than in Oct. The higher level of DIN in May and Aug may be related to terrestrial input and supply for the demand of phytoplankton growth. In October, the temperature and DIN content were both not suitable for phytoplankton growth, and Chl_a showed the lowest value. Spatially, the DIN distribution across the three seasons was rather similar, with high values observed in nearshore waters and low values in offshore waters (Dataset S1 & Fig. S1), which suggested that terrestrial input was an important source of DIN. The pH, soluble reactive phosphate (SRP) and chemical oxygen demand (COD) showed relatively higher values in October than in August and May, which may be caused by the dead phytoplankton release and terrestrial loadings through coasts and rivers. The dissolved oxygen (DO), conductivity and depth did not show significant variation among sampling times (Table 1), while the conductivity and depth had relatively higher values at offshore stations (Dataset S1) since the more remote the sampling water was, the greater the depth was in Bohai Bay and the closer it was to the open sea with higher salinity and conductivity. The ordination plot showed distinct partitioning of samples from nearshore and offshore sites along principal component axis 1 (PC1) (Fig. 1). The ordination plot could explain 73.49% of the total variation in the geo-physical–chemical parameters and revealed a linear positive correlation between different parameters (Fig. 1). AN, DIN, nitrate and Chl_a were most crucial in the partitioning of samples from May and the other 2 months; salinity, longitude, depth and conductivity were crucial for the partitioning of samples from offshore and nearshore stations; pH, COD, SRP, nitrite and temperature were crucial for the partitioning of samples from nearshore stations in August and October and samples from offshore stations. Overall, the principal component analysis (PCA) plot clearly showed both the temporal and spatial variation of the measured environmental parameters, indicating that complex biogeochemical processes and hydrodynamic conditions lead to the variation among sites and seasons.Table 1 The independent-samples t test of environmental variables and α-diversity among different months.Full size tableFigure 1Biplot of the principal component analysis (PCA) for environmental parameters in the seawater samples of the Bohai Bay coastal area across different seasons and sites. The two principal components (PC1 and PC2) explained 73.49% of the total variation in the environmental data and showed clear partitioning of offshore samples (in blue font) from other nearshore samples along PC1 and partitioning of May samples from August and October along PC2. The variables transparency and latitude were strongly correlated with PC1, and the variables ammonia nitrogen (AN), COD, pH, soluble reactive phosphate (SRP), and nitrite were strongly correlated with PC2. Chlorophyll a (Chl_a), dissolved inorganic nitrogen (DIN), nitrate and DO were mainly positively correlated with samples from May, while salinity, longitude, depth and conductivity were mainly positively correlated with offshore samples. Blue arrows represent environmental parameters, and circles in color represent sampling points.Full size imageProkaryotic α/β-diversity variationMeasures of α-diversity showed significant differences in shannon, evenness, faith_pd and OTU richness between samples from May/Aug and Oct (Fig. 2, Table 1). Principal coordinates analysis (PCoAs) based on weighted UniFrac (WUF) distance and unweighted UniFrac (UUF) distance showed that the PCC from different sampling months separated across the first and second principal coordinates (Fig. 3A-B). Both the analysis of similarity (ANOSIM) and permutational multivariate analysis of variance (PERMANOVA/ADONIS) results indicated that the prokaryotic communities varied significantly across different sampling months when using a WUF distance metric (ANOSIM, r = 0.709, P = 0.001; ADONIS, R2 = 40.0%, P = 0.001) and UUF distance metric (ANOSIM, r = 0.934, P = 0.001; ADONIS, R2 = 38.7%, P = 0.001). At the same time, the prokaryotic α– and β-diversity both showed high within-month variability in Aug (Figs. 2, 3C–D), which indicated that the community varied greatly among different sites in Aug.Figure 2Alpha diversity of shannon, eveness, faith_pd (phylogenetic diversity) and OTU richness value of the prokaryotic community of all the samples from different stations at different sampling times.Full size imageFigure 3Principal coordinate analysis (PCoA) based on unweighted (A) and weighted (B) UniFrac distances for prokaryotic communities in the surface waters; box plots showing the unweighted (C) and weighted (D) UniFrac distances among each station at different sampling times.Full size imageCorrelation between prokaryotic α/β-diversity and physical, chemical and geographic factorsThe α-diversity measurements exhibited significant positive correlations with temperature, pH, SRP, AN and un_ionN (Dataset S2). The correlation between α-diversity indexes and geo factors (longitude and latitude) was not strong or significant both in samples across the three sampling times or from each sampling time (Dataset S2).The environmental variation significantly correlated with β-diversity among the three seasons (r_weighted = 0.4558, r_unweighted = 0.4631, P = 0.001, Table 2), with pH, AN, temperature, un_ionN, COD, nitrite, SRP, salinity, DO and DIN as the main individual determinants. However, it did not show significant correlations with β-diversity at any sampling time except in Oct (Table S1).Table 2 Spearman’s rank correlation between environmental/spatial variability (Euclidean distance) and prokaryotic β-diversity (weighted/unweighted UniFrac distance) among all samples from different season.Full size tableThe geographic distance was not correlated with prokaryotic β-diversity (variation in community composition; r  0.05; Table 2) among the three sampling times. However, samples from Aug and Oct exhibited a significant correlation between β-diversity and geographic distance (Table S1).Factors driving the PCC variationPERMANOVA using the UUF/WUF distance indicated that temperature variation explained the largest part of community variation among the investigated factors (34.90%/19.83%, P = 0.001, Dataset S3), with AN (31.84%/13.56%, P = 0.001) and salinity (12.91%/6.21%, P = 0.001) as the second and third most significant sources of variation.The variance partitioning analysis (VPA) conducted on both UUF/WUF distances showed that almost 100% percent of the variation in PCC among all three sampling times was explained by the detected environmental factors. In May, no environmental or spatial factors could be selected as significantly explain the PCC variation; in Aug, the joint effects of environmental and spatial factors could explain 49.5% of the variation; in Oct, based on WUF distance, the spatial factors could purely explain 10.5%, environmental factors could purely explain 38.8%, their joint effects could explain 28.2%, and based on UUF distance, the joint effects of environmental factors and trend could explain 13.7% of the PCC variation. These results indicated dramatic shifts in the spatial or environmental factor effects on the PCC variation at different sampling times in Bohai Bay (Table 3).Table 3 Variance partitioning analysis of prokaryotic community in Bohai Bay according to seawater environmental factors and geospatial factors. The spatial factors including linear trend and PCNM variables. Forward selection procedures were used to select the best subset of environmental, trend, and PCNM variables explaining community variation, respectively. The community variation was calculated on the weighted and unweighted UniFrac distance matrix, respectively. Monte Carlo permutation test was performed on each set without the effect of the other by permuting samples freely (999 permutations).Full size tableDistinct seasonal features at the phylum and OTU levelsThere were notable differences in the proportions of various phyla among different seasons (sampling month). In May, there was a greater proportion of Alphaproteobacteria (41.41%), Planctomycetes (6.42%), Actinobacteria (3.86%), Firmicutes (1.48%), Acidobacteria (0.45%), TM7 (0.16%), Tenericutes (0.16%), OD1 (0.13%), and WPS-2 (0.09%) than in Aug and Oct, whereas Gammaproteobacteria (44.23%), GN02 (0.08%) and SAR406 (0.04%) were depleted in May and Aug but enriched in Oct. In Aug, Bacteroidetes (13.98%), Deltaproteobacteria (6.93%), Verrucomicrobia (4.5%), Chloroflexi (0.36%), Lentisphaerae (0.97%), TM6 (0.25%), Nitrospirae (0.08%), Chlamydiae (0.07%), Chlorobi (0.07%), Spirochaetes (0.04%) and OP8 (0.03%) were significantly enriched than in the other two sampling times (Duncan test; Table S2).At the OTU level, OTUs with relative abundance  > 0.01% (1040 OTUs) were used to perform the difference analysis, and 175 OTUs in May, 281 OTUs in Aug, and 210 OTUs in Oct were identified as seasonal specific OTUs (ssOTUs). The cooccurrence network showed that the ssOTUs were clustered separately (Fig. 4A). Furthermore, the separation of the three modules contained most of the ssOTUs specific to different seasons (Fig. 4A-B). All the ssOTUs of different seasons comprised a taxonomically broad set of prokaryotes at the phylum (phylum Proteobacteria is grouped at the class level) level (Fig. 4C) belonging to various phyla with different proportions. Betaproteobacteria, Verrucomicrobia, Gemmatimonadetes, Epsilonproteobacteria, PAUC34f., and Euryarchaeota did not show significant differences among the three sampling times at the phylum level, but features belonging to these phyla showed differences at the OTU level (Fig. 4C, Dataset S4). In addition, the phylum ssOTUs belonging to, such as Alphaproteobacteria, Gammaproteobacteria, Bacteroidetes, Actinobacteria, and Deltaproteobacteria, were not only enriched at one sampling time (Dataset S4) but also enriched at the other two sampling times (Fig. 4C, Dataset S4). These results revealed that different seasons do not strictly select specific microbial lineages at the phylum level, but a finer level analysis could more strictly reflect the seasonal variation.Figure 4Co-occurrence patterns of seasonal sensitive OTUs (A). Co-occurrence network visualizing significant correlations (ρ  > 0.7, P  0.01%. Different colors represent ssOTUs in May (green), Aug (red) and Oct (blue). Cumulative relative abundance (as counts per million, CPM; y-axis in × 1000) of all the sensitive modules in the networks (B). The phylum attribution of ssOTUs in each season (C). The y-axis is the percentage of the number of OTUs that belong to a particular phylum that accounts for the total number of all the OTUs.Full size imageRegression analysis between the relative abundance of modules to which the ssOTUs belonged and the environmental factors was also conducted, and module 1 abundance, to which the Aug-ssOTUs belonged, showed a significant positive correlation with temperature (R2 = 0.77, P = 6.609e−62), AN (R2 = 0.43, P = 7.416e−25), and un_ionN (R2 = 0.75, P = 1.366e−58) and a negative correlation with SRP (R2 = 0.81, P = 6.762e-17). This may be caused by the functional role of the microbes in Aug. In the Aug-ssOTUs, Deltaproteobacteria showed a higher ratio than in the other 2 months (Fig. 4c), and in the following functional analysis, Deltaproteobacteria contributed to the genes related to nitrogen fixation, which may help to explain why there was a positive correlation of Aug-ssOTUs to AN and un_ionN. The module 2 abundance to which the May-ssOTUs belonged showed a significant negative correlation with pH (R2 = 0.65, P = 4.026e−44), temperature (R2 = 0.19, P = 2.325e−10), un_ionN (R2 = 0.025, P = 0.01779), and SRP (R2 = 0.12, P = 4.104e−07) and a positive correlation with AN (R2 = 0.26, P = 5.174e−14). In the May-ssOTUs, the ratio of Alphaproteobacteria was the highest, and Alphaproteobacteria were reported to be pH-sensitive groups in marine environments20, which prefer neutral pH environments21. In this study, the pH in May was 8.04 ± 0.07, in Aug was 8.39 ± 0.09, in Oct was 8.38 ± 0.07, and the pH in May was the closest to neutral, and the ratio decreased with increasing pH in Oct and Aug. The abundance of module 3, to which the Oct-ssOTUs belonged, showed a significant positive correlation with SRP (R2 = 0.81, P = 0.16e-10) and pH (R2 = 0.054, P = 0.00075) and a negative correlation with temperature (R2 = 0.44, P = 2.276e−25), AN (R2 = 0.75, P = 4.51e−58), and un_ionN (R2 = 0.6, P = 3.995e-39) (Fig. S2). Phosphate has been identified to limit primary productivity22, which is of great importance in the structure of dominant bacterial taxa in marine environments23. In the Oct-ssOTUs, the ratio of Gammaproteobacteria was the highest, as reported. Gammaproteobacteria was significantly explained by SRP during the seasonal variation in the Western English Channel, with Rho equal to 0.7523, which suggested the sensitivity of it to SRP, and in that study, it also showed a negative correlation between temperature and Gammaproteobacteria and a positive correlation between SRP and Gammaproteobacteria. Although the correlation was not significant, the variation trend was consistent, which indicates that the phenomenon observed in this study was not unexpected. In addition, most ammonia-oxidizing bacteria belong to the Betaproteobacteria and Gammaproteobacteria classes are chemolithoautotrophs that oxidize ammonia to nitrite24. Gammaproteobacteria and Betaproteobacteria both had higher ratios in Oct-ssOTUs, and the functional prediction results also showed that pmoA/amoA and pmoB/amoB, which encode ammonia monooxygenase, were mainly contributed by OTUs from Gammaproteobacteria and Betaproteobacteria (Dataset S10). The utilization of ammonia may explain the negative correlation between the Oct-ssOTUs and AN.Community assembly processes across different sampling months and sitesBased on the analysis of phylogenetic turnover, unweighted βNTI mostly ranged from -2 to 2 across different sites at a single sampling time in May, Aug and Oct, revealing that PCC variations across different sampling sites at a single time were mostly affected by stochastic processes. The unweighted βNTI was greater than 2 across May–Aug, May–Oct and Aug-Oct (Fig. 5A), which revealed that the variations in PCC across different sampling times were mostly affected by deterministic processes. The RCbray values across any two sampling times were equal to 1, and in each sampling time, the RCbray values ranged from − 1 to 1 (Fig. 5B). Combining the βNTI and RCbray values, the community assembly processes were quantified at each sampling time and at any two sampling times. As shown in Fig. 5C, turning over of the community during different sampling times was mainly governed by selection; among the different sites in May and Oct, it was mainly governed by “undominated” processes; community turn over in Aug was mainly governed by the influence of “Dispersal Limitation”. These results indicated that the shifts in the assembly of prokaryotic communities during different sampling times were caused by strong “heterogeneous selection” (βNTI  > 2), and the community variation at each sampling time was mainly caused by stochastic processes.Figure 5Patterns of distribution of unweighted βNTI (A) and RCbray (B) values across different sampling times. Quantification of the features that impose community assembly processes in and among different sampling times. (C) Pie charts give the percent of turnover in community composition governed primarily by Selection acting alone (white fill), Dispersal Limitation (green line fill), Homogenizing Dispersal (blue line fill) and undominated process (cyan fill).Full size imagePrediction of the metabolic potential at different sampling timesThe NSTI scores of each sample ranged from 0.033 to 0.096, with a mean of 0.058 (Dataset S5). Microbial functions were detected in all the samples from the three sampling times, and it was found that the relative abundances of 242 pathways were significantly changed between samples from May and samples from Aug (Dataset S6). The relative abundances of 321 pathways were significantly changed between samples from May and Oct (Dataset S7), and the relative abundances of 370 pathways were significantly changed between samples from Aug and Oct (Dataset S8).Genes related to energy metabolism were given more attention. For nitrogen metabolism genes relevant with nitrogen fixation (nifD, nifK) were detected only enriched in Aug, while genes relevant with nitrate reduction and denitrification (narG, narZ, nxrA, narH, narY, nxrB, narI, narV, nirD, nasA, nasB) were detected enriched in May, genes related with ammonia oxidation were both detected enriched in Oct and Aug. For sulfur metabolism, genes relevant with thiosulfate oxidation (soxA, soxB, soxC, soxX, soxY and soxZ) were only enriched in Aug, while genes relevant with sulfate and sulfite reduction (cysNC, aprA, aprB, cysJ, cysI, cysK, dsrA) were detected enriched in May and Oct (Fig. 6).Figure 6The LEfSe analysis indicated significantly differential abundances of PICRUSt predicted genes relevant to energy metabolism in different months of samples.Full size imageProkaryotic taxa contributed to the significantly varied functional genes related to nitrogen and sulfur metabolism at different sampling times. At the species level, the taxa contributing to nifK and nifD mainly belonged to Deltaproteobacteria and Firmicutes, and the taxa contributing to the sox-series genes mainly belonged to Alphaproteobacteria and Gammaproteobacteria (Fig. S3). The denitrification-related functional genes that were enriched in May were mainly contributed by members from Alphaproteobacteria and Gammaproteobacteria. The taxa contributing to dsrA, aprA and aprB were mainly from Deltaproteobacteria, including members of Desulfarculaceae, Desulfobacteraceae, Desulfobulbaceae, Desulfovibrionaceae and Syntrophobacteraceae (Fig. S4). More

Study sites and coloniesAll the experiments were carried out during the apple and pear blooming seasons of 2007, 2008, 2011, 2013 and 2014 in different locations of the province of Rio Negro, Argentina, while some laboratory experiments performed in the city of Buenos Aires. We used individual foragers of Apis mellifera L. and their colonies containing a mated queen, brood, and food reserves in ten-frame Langstroth hives. All beehives used had similar sizes and the same management history from the beekeeper. The honey bees studied belonged to commercial Langstroth-type hives rented to pollinate these plots. Each hive had a fertilized queen, 3 or 4 capped brood frames, reserves and approximately 15,000 individuals56.Testing generalization of memories from pear mimic odours to pear and apple natural floral scentsThe absolute conditioning assays were performed in the laboratories of the School of Exacts and Natural Sciences of the University of Buenos Aires (34° 32′ S, 58° 26′ W), Buenos Aires, Argentina. We used honey bee foragers collected at the entrance of the hives settle in the experimental field of the School of Exacts and Natural Sciences. The apple (‘Granny Smith’ and ‘Red Delicious’ varieties) and pear (‘Packham’ and ‘D’anjou’ varieties) bud samples that we used as conditioned stimuli (CS) during the conditioning were collected at the end of the blossom of 2011 in Ingeniero Huergo (39° 03′ 27.5″ S; 67° 13′ 53.5″ W), province of Río Negro, Argentina, and taken to the laboratory in the city of Buenos Aires, Argentina, to be used within the following 2 days.We first developed the three different synthetic mixtures (PM, PMI and PMII) that could be generalized to the fragrance of the pear flower by foraging bees. The pear synthetic mixtures were formulated considering the previously reported volatile profile of pear blossoms57. Then, we chose the synthetic mixture most perceptually similar to the pear flower fragrance and measured its generalisation response to the apple flower fragrance to test the compounds’ specificity. The chemical compounds used to prepare the different synthetic mixtures for the behavioural assays were obtained from Sigma-Aldrich, Steinheim, Germany. The compounds used for the three pear mixtures (PM, PMI and PMII) were composed by alpha-pinene, 2-ethyl-hexanol, (R)-(+)-limonene, and (±)-linalool. For details of the PM and mixture proportions see Patent PCT/IB2018/05555058.To test generalization, we took advantages of the fact that honey bees reflexively extend their proboscises when sugar solution is applied to their antennae59. The proboscis extension reflex (PER) can be used to condition bees to an odour if a neutral olfactory stimulus (CS) is paired with a sucrose reward as unconditioned stimulus, US60. Conditioned honey bees extend their proboscises towards the odour alone, a response that indicates that this stimulus has been learned and predicts the oncoming food reward. Conditioned bees can generalize such a learned response to a novel odour if it is perceived like the conditioned one (CS). Then we performed three absolute PER conditionings where we paired each of the three PMs with a sucrose-water solution (30%) reward along three learning trials (exp. 4.2a). Afterwards, pear floral scent was presented as novel odour to test generalization. Based on the generalization level to the pear odour, we chose the synthetic mixture that showed the highest generalisation towards pear flower fragrance, and we used it in all the experiments that follow. In an additional 3-trial PER conditioning with the chosen mixture, we quantified generalisation towards both the pear and apple fragrances as novel stimuli (exp. 4.2b).The experimental bees were all foragers, captured from colonies that had no access to any pear and/or apple tree, hence completely naïve for the CSs. Immediately after capture, bees were anaesthetized at 4 °C and harnessed in metal tubes so that they could only move their mouthparts and antennae60. They were fed 30% weight/weight unscented sucrose solution for about three seconds and kept in a dark incubator (30 °C, 55% relative humidity) for about two hours. Only those bees that showed the unconditioned response (the reflexive extension of the proboscis after applying a 30% w/w sucrose solution to the antennae) and did not respond to the mechanical air flow stimulus were used. Trials lasted 46 s and presented three steps: 20 s of clean air, 6 s of odour presentation (CS) and the last 20 s of clean air. During rewarded trials (CS), the reward (US, a drop of 30% w/w sucrose solution) was delivered upon the last 3 s of CS presentation. The synthetic mixtures (PM) were delivered in a constant air flow (15 ml/s) that passed through a 1 ml syringe containing 4 µl of the synthetic mixture on a small strip of filter paper. On the other hand, pear and apple floral volatiles were swept from a 100 g of fresh pear buds (var. ‘D’Anjou’ and ‘Packham’) or apple buds (var. ‘Granny Smith’, ‘Gala’ and ‘Red Delicious’) inside a kitasato by means of an air flow (54 ml/s).Testing discrimination between mimics and natural floral scentsThe differential conditioning assays were performed in a field laboratory in Ingeniero Huergo, province of Río Negro, Argentina. Conditioning trials with AM as CS were carried out in September 2007 and 2008, prior to the beginning of flowering of the fruit trees. Conditioning trials with PM as CS were carried out in September 2011 in the same area (Ingeniero Huergo, province of Río Negro, Argentina). Apple and pear bud samples used as CS were collected in plots that start blooming located around Ingeniero Huergo, but distant (more than 1 km) from the plot where we collected the bees. The bud samples presented the following varieties: M. domesticus sp., ‘Granny Smith’, ‘Gala’, and ‘Red Delicious’; P. communis sp., ‘Packham’ and ‘D’Anjou’.With the aim to develop a synthetic mixture that presents difficult to discriminate with the fragrance of the apple flower by foraging bees, an apple synthetic mixture (AM) was formulated considering the previously reported volatile profile of apple blossoms61. The chemical compounds used to prepare the apple synthetic mixtures for the behavioural assays were obtained from Sigma-Aldrich, Steinheim, Germany. Apple mimic (AM) was composed by benzaldehyde, limonene and citral. For details of the AM proportions see Patent AR2011010244162. Jasmine mimic (JM) was a commercial extract obtained from Firmenich S.A.I.C. y F, Argentina.If the synthetic mixture chosen were perceptually similar to the apple flower fragrance, experimental bees should have difficult to discriminate to the apple flower fragrance to test the compounds’ specificity. Thus, we performed differential PER conditioning between synthetic mixtures (AM and Jasmine mimic, JM) or between synthetic mixtures (AM or JM) and the apple natural fragrance. We followed a differential PER conditioning34 to assess to what extent the bees were able to discriminate the synthetic mimics from their natural flower scents. PER differential conditioning consisted of four pairs of trials, four rewarded trials (CS+) and four non-rewarded trials (CS−) that were presented in a pseudo-randomized manner. Conditionings were performed using the synthetic mixtures PM and AM and the natural floral scents, pear and apple, either as CS+ and CS−. We followed the same procedure that in 3.3 to capture the bees and to present the stimuli during trials.Feeding protocolWe used the offering of scented sucrose solution in the hive as a standardized procedure to establish long-term olfactory memory in honey bees23,24,24,26,63. Scented sucrose solution was obtained by diluting 50 µl of PM or AM per litre of sucrose solution (50% weight/weight, henceforth: w/w). For the ‘apple’ series, colonies were fed 1500 ml of sugar solution offered in an internal plastic feeder for 2 days, about 3 days before the apple trees began to bloom. For the ‘pear’ series, hives were fed 500 ml of sugar solution that we spread over the top of the central frames. Both feeding procedures have been found to be functional for establishing olfactory in-hive memories26. Depending on the pear varieties, the scented sucrose solution was offered when the pear trees were 10–40% in bloom.Colony activityThe effects of the AM-treatment on colony nest entrance activity were studied in 18 colonies located in an agricultural setting of apple and pear trees in Ingeniero Huergo, on an 8-ha plot, half of which was planted with apple trees (varieties: ‘Granny Smith’, ‘Gala’ and ‘Red Delicious’) and the other 4 ha with pear trees (varieties: ‘Packham’ and ‘D’anjou’). The effect of the PM-treatment on colony activity was studied in 14 colonies located in three adjoining pear plots (total surface: 8 ha) in Otto Krause (39° 06′ 22″ S 66° 59′ 46″ O, Supplementary Fig. S5), province of Río Negro, Argentina. The varieties of these plots corresponded to ‘Packham’ and ‘Williams’. Pollen collection (exp. 4.5.2) was also studied in colonies located in these plots.We focused on the nest entrance activity since once the first successful foragers return to the hive and display dances and/or unload the food collected, it promotes the activation or reactivation of inactive foragers and, in a minor proportion, those hive mates ready to initiate foraging tasks39,65,66,67,67. Then, we choose number of incoming bees as an indicator of colony foraging activity, since most of these bees are expected to return from foraging sites33. Thus, we compared the activity level at the nest entrance between 7 SS + PM-treated colonies and 7 SS-treated colonies. We also compared the nest entrance activity level between 5 colonies treated with SS + AM and 5 colonies fed with SS. This activity value was estimated by the number of incoming foragers at the entrance of the hive for one minute, every morning at the same time (10:30 a.m.) during the entire experiment (9 consecutive days). A first measurement was done one day before feeding the colonies (used as covariate) and 7 measurements afterwards.We measured the amount of pollen loads collected by two colonies: one fed with SS + PM and one fed with SS. Pollen loads were collected using conventional pollen traps (frontal-entrance trap), consisting of a wooden structure with a removable metal mesh inside. Pollen samples were collected for 3 days, two hours per day during the late morning, 3, 7 and 8 days after the offering of SS + PM or SS. Pollen pellets identified based on pollen colour as coming from the pear flower or from other species were separated and counted. In addition, we estimated the weight of pear pollen loads during a 5 days period, from 6 to 10 days after the offering of scented or unscented sucrose solution. To reduce measurement error, pollen loads were weighed in groups of 10.Crop yieldPear crop yield was studied in pear plots in General Roca (39° 02′ 00″ S; 67° 35′ 00″ O, Supplementary Fig. S4, Supplementary Table S3), province of Río Negro, Argentina. In an area of 15.2 ha (4 plots of 3.8 ha each), 45 beehives were equidistantly located in groups. We measured the number of fruits per tree set of 30 trees in the surrounding areas of the PM-treated colonies (2 groups of 8 hives) and control colonies (2 groups of 8 hives). A third group category contained 13 untreated colonies. The varieties of the pear trees were ‘D’Anjou’ and ‘Packham’.Apple crop yield estimated by means of number of fruits per plant was studied in General Roca (Supplementary Fig. S2, Supplementary Table S1), province of Río Negro, Argentina. We measured fruit set in the two plots that covered a surface of 3.8 ha and contained a total of 74 colonies distributed in groups (the control plot, 39 SS-treated-colonies treated with SS; and the treated plot, 35 SS + AM-treated-colonies treated with SS + AM). The varieties of the apple trees were ‘Red Delicious’ (clone 1), ‘Royal Gala’ and ‘Granny Smith’.A second studied on apple fruit yield by means of kg of fruits per hectare was performed in Coronel Belisle (39° 11′ 00″ S 65° 59′ 00″ O, Supplementary Fig. S3, Supplementary Table S2), province of Río Negro, Argentina. Four apple plots with ‘Granny Smith’, ‘Hi Early’ and ‘Red Delicious’, clone 1 varieties of 15.4 ha each were randomly assigned to different treatments (treated plot 1, 40 SS + AM-treated-hives treated with SS + AM; treated plot 2, 40 SS + AM-treated-hives treated with SS + AM; control plot 1, 40 SS-treated-hives treated with SS; control plot 2, 40 SS-treated-hives treated with SS).During the fruit harvest, the fruit yield was estimated in the surroundings (150 m around) of two groups of 8 colonies each. We fed one group SS + PM and the other unscented sucrose solution (SS). Yield was estimated as the number of fruits per trees in 30 randomly selected trees within each area, alternating the counts between the North and South faces of the plots. Following the same procedure, we also estimated the number of fruits per trees in the surroundings of two groups of 14 colonies each that pollinated apple crops. Again, we fed one group SS + AM and the other SS. Additionally, a total of 218 colonies in General Roca and 180 colonies in Coronel Belisle have been separated in the two experimental groups, in which yield had been provided by the producer and expressed in kg of fruits per ha. It is worth remarking that in some plots the distance between treated and control beehive groups was around 300 m, suggesting that might have been overlapping flying areas between treated and control hives. Additionally, the apple fields studied in the surrounding of Coronel Belisle, presented many trees without flowers. It was considered that the absence of flowers in numerous trees would bias the counts performed in those fields. Then, to quantify this situation, which might be associated with the masting phenomenon68, samples with the proportions of trees without flowers for every 20 trees in each plot was done. Trees that had between 80 and 100% of their surface devoid of flowers were considered “without flowers” trees, and “trees with available flowers” those that had more than 20% of their surface covered with flowers. An average of 30% of the trees within these plots were devoid of flowers. Thus, a correction factor was considered to evaluate the yield data provided by the grower per plot analysed (Supplementary Table S4).StatisticsAll statistical analyses were performed with R Core Team 201969. For Experiment 4.2 and 4.3, we analysed PER proportion by means of a binomial multiplicative generalized linear mixed model using the “glmer” function of the ‘lme4’ package70.For experiment 4.2a we considered the pear mimics (three-level factor corresponding to PM, PMI and PMII) and the event (two-level factor corresponding to 3rd trial and test) as fixed factors and each “bee” as a random factor.For experiment 4.2b we considered the tested odours (three-level factor corresponding to Apple, Pear and PM) as fixed factors.For experiment 4.3 we considered the tested odours (two-level factor corresponding to CS+ and CS−) as fixed factors. Post hoc contrasts were conducted on models to assess effects and significance between fixed factors using the “emmeans” function of the ‘emmeans’ package version 1.7.071 with a significance level of 0.05.For experiment 4.5.1 we analysed “rate of incoming bees” using a generalized linear mixed model. As Poisson model for incoming bees was overdispersed72, we used a negative binomial distribution using the ‘glmmTMB’ package (function ‘glmmTMB’73. We considered “treatment” [two-level factor corresponding to SS + AM (or SS + PM) and SS], “days” (7-level factor corresponding to the date after treatment), the rate of incoming bees before the offering of food (to control for pre-existing colony differences) as covariate (a quantitative fixed effects variable), and “colony” as a random factor.For experiment 4.6, we analysed fruits per trees by means of a negative binomial multiplicative generalized linear mixed model using the “log” function of the ‘ml’ package70. Post hoc contrasts were conducted on models to assess effects and significance between fixed factors using the “emmeans” function of the ‘emmeans’ package version 1.8.071 with a significance level of 0.05. For experiment 4.6b we analysed “yield” (as weight of fruits per unit area) using a general linear mixed model. We checked homoscedasticity and normality assumptions (Levene and Shapiro–Wilk tests, respectively). We considered “treatment” (two-level factor corresponding to SS + AM and SS) and “apple varieties” (3-level factor corresponding to Hi Early, Granny Smith and Chañar 28) as fixed factors and “location” (2-level factor corresponding to General Roca and Coronel Belisle) as random factors. More

Population sampling and sequencingWe generated whole genome sequence data from 302 wild-caught individual An. minimus female mosquitoes collected from five different field sites in Cambodia using the Illumina HiSeq 2000 platform with 150 bp paired-end reads with a target coverage of 30X for each. Mosquito collections in Thmar Da, in Eastern Cambodia, were done in 2010. Longitudinal monthly collections were performed from February 2014 to January 2015 in two sites in each of the Preah Vihear, and Ratanakiri provinces. Quarterly collections were also done in 2016 in one site in Preah Vihear province, Cambodia.Variant discoveryThe methods for sequencing and variant calling closely follow those of the Anopheles gambiae 1000 Genomes project phase 2 (Ag1000G)27. Sequence reads were aligned to the An. minimus reference genome AminM128. We restricted our analysis to the largest 40 contigs, which cover 96.6% of the AminM1 reference genome, as many smaller-sized contigs can confound diversity and divergence calculations. We found that 138,161,075 (75.4%) of sites within these 40 largest contigs pass our site filters and thus were accessible to SNP calling. Of these, we discovered 38,000,285 segregating single nucleotide polymorphisms (SNPs) that passed all of our quality control filters of 55,307,039 total segregating SNPs. 13.4% of these SNPs were multiallelic, with 32,906,471 biallelic SNPs. There were 4,807,355 triallelic and 286,459 quadriallelic SNPs. A total of 100,160,790 sites were invariant. The median genome-wide coverage was 35X.Population structureA principal component analysis (PCA) over biallelic SNPs distributed over the genome of 302 individual field-collected mosquitoes showed that there is clear population structure of An. minimus in Cambodia. Samples collected from five sites in three provinces split into three distinct clusters; here, we report on 283 individuals that could be clearly assigned to these clusters (Fig. 1), excluding 9 anomalous and 10 outlying individuals. One cluster includes all samples from the western collection site Thmar Da and the northern collection sites in Preah Vihear province, with two further clusters with samples from Ratanakiri province in the northeast. These clusters split primarily along the first and second principal components. This was a surprising finding because this population structure did not correlate to the geographic sampling of these mosquitoes. Individuals collected from the western and northern sites cluster tightly together despite being hundreds of kilometers apart.Fig. 1: Population structure of An. minimus in Cambodia.The map indicates the five Cambodian collection sites. Principal component analysis (PCA) of whole genome sequences of 283 individual An. minimus s.s. collected in five villages in Cambodia shows that there is a distinct population structure and three populations. When performing the same PCA on a large X-chromosomal contig (KB664054), these individuals break into four populations: TD from the West, PV from the northern province in Preah Vihear, and RK1 and RK2, both collected in two sites in Ratanakiri province in the Northeast.Full size imageTo further explore this population structure, we performed the same PCA over individual contigs from different regions of the genome. Performing PCA over the largest X-chromosomal contig KB664054 resulted in a splitting of the western and northern samples, indicating four distinct populations of An. minimus in Cambodia (Fig. 1). PCA from this contig on a quickly evolving sex chromosome revealed more population structure compared to autosomal contigs. The populations defined by these PCA clusters are designated in this study as TD from Thmar Da, in Western Cambodia (n = 41), on the Thai-Cambodian border, PV from the Northern province Preah Vihear (n = 156), and the two distinct populations collected in Ratanakiri province in the Northeast, each including individuals collected at both collection sites, these are designated as populations RK1 (n = 58) and RK2 (n = 28).To confirm our results from PCA, we also performed an admixture analysis. We ran admixture on each of the largest 10 contigs for values of K between 2 and 6 (Supplemental Fig. 1). At K = 2, the samples from Northeastern Cambodia split from Northern and Western Cambodia samples. At K = 3, the two different groups in Ratanakiri were separated, consistent with the PCA results. At K = 4, there was some evidence for geographical population structure between the Western TD and Northern PV populations, but the admixture results did not perfectly correspond with geographic sampling, with some evidence of mixed ancestry in the PV samples. Again, this is consistent with the PCA groupings, with the generally weaker evidence of geographic population structure between TD and PV. A cross-validation analysis showed the lowest cross-validation error for K = 2 and K = 3, consistent with the strongest evidence for population structure between the two RK groups and other populations. Cross-validation error was higher at K = 4, consistent with the weaker differentiation between TD and PV. At no point was their an indication of admixture between RK1 and RK2.To examine population differentiation, we computed differences in allele frequencies between each population using Pairwise Fst. Pairwise Fst between all 4 populations over the largest contig, KB663610, representing 16% of the An. minimus genome, (Fig. 2) shows that differentiation was relatively low between populations of TD and PV with an average pairwise Fst of 0.003, while the difference between RK2 and the other three populations is tenfold higher, around 0.03. Pairwise Fst estimates comparing these populations over other large An. minimus contigs indicate a similar level of differentiation, with average pairwise Fst values over 0.03 (Supplementary Data 3). The two sympatric populations from the Ratanakiri collection sites are as differentiated from each other as they are from the northern and western clusters.Fig. 2: Population diversity and divergence.Nucleotide diversity (π), Watterson’s Theta (θW), and Tajima’s D statistics were calculated over fourfold degenerate sites on autosomal contigs. The error bars indicate 95% confidence intervals calculated over 100 bootstrap replicates over samples. An average pairwise Fst in the table here was calculated in 20 kb windows over the largest contig KB663610.Full size imageThis level of differentiation of RK2, even from the RK1 population, might indicate an emerging cryptic species within An. minimus A or a newly diverging clade. RK1 and RK2 are sympatric populations, both being collected in the same two sites in Northeastern Cambodia. The differences seen here between RK1 and RK2 populations are consistent with cryptic taxa in other anopheline groups. For example, in the An. gambiae complex, the level of differentiation between recently diverged sibling species An. coluzzii and An. gambiae in West Africa is approximately 0.0319.Population diversity and variationTo characterize population diversity among these populations, nucleotide diversity (π), Watterson’s Theta (θW), and Tajima’s D statistics were calculated over 4-fold degenerate sites on autosomal contigs larger than 2 megabases with 100 bootstrap replicates over samples. These 17 contigs represent 80% of the Anopheles minimus genome (Fig. 2). The populations were downsampled for these calculations to have sizes equal to that of the smallest population RK2 (n = 28).There are small but significant differences in the magnitude of the genetic diversity summary statistics between these four different populations. In particular, there were notable differences between the putatively cryptic taxa RK1 and RK2, two populations that were collected in the same sites in Northeastern Cambodia. RK1 had higher levels of nucleotide diversity and lower levels of Tajima’s D than RK2. These differences are consistent with different population size histories between these sympatric groups. Lower values of Tajima’s D suggest stronger population growth in RK1. Comparing all four populations, higher levels of genetic diversity indicate larger effective population sizes of TD and PV compared to RK1 and RK2.RK2 has a significantly reduced nucleotide diversity and Watterson’s Theta compared to the other three populations. This may indicate a smaller population size and a recent bottleneck of the RK2 population in Cambodia. All four An. minimus populations have a negative Tajima’s D, indicating an excess of rare variants, particularly in RK1. This suggests recent population expansions in all populations.Signals of evolutionary selectionWe used Fst to scan across the Anopheles minimus genome to look for regions of the genome with increased differentiation. When we scanned the genome using pairwise Fst, there were no apparent long signals of differentiation that might indicate a large inversion or other structural variants, known to be major drivers of adaptive evolution in other Anopheles groups. To investigate increased differentiation across large regions of the genome, we performed scans of nucleotide diversity (π), Watterson’s Theta (θW), and Tajima’s D over the largest 14 contigs (representing 80% of the An. minimus genome). As with the Fst scans, there were no large regions of higher differentiation in any of the populations that might indicate major structural variants or inversions (Supplementary Figs. 2–4).Whole-genome sequencing allowed us to identify pointed signals occurring across the entire genome using scans of average pairwise Fst. Isolated points of high differentiation were compared over single contigs with average pairwise Fst calculated over windows of 1000 SNPs each and plotted over whole contigs. The strongest signals, indicated by the highest Fst value at the peak of a strong signal of differentiation, were ranked and compared. The five top signals in each of the six comparisons between the four populations are listed in Table 1. These isolated points of high differentiation are one indication of a signal of evolutionary selection. The most differentiated regions by Fst occurred when comparing the RK2 population to the other three populations, with the highest selection peaks with pairwise Fst over 0.4. RK2 also had more distinct signals of selection when compared to the other populations than RK1. Since these signals of differentiation were highly localized, we could look to known gene annotations and gene predictions across the AminM1 reference genome to see which genes were within 100 kbp of the peaks of these signals. We have noted candidate genes of interest that were near the strongest Fst signal peaks and also had known or predicted gene functions (Table 1, Supplementary Fig. 6, Supplementary Fig. 8).Table 1 The top five Fst signals of high differentiation within each of six population comparisons are reported here.Full size tableThere is almost no indication of selection when comparing the Thmar Da population with Preah Vihear, with all but one signal with Fst values below 0.05. The one strong signal between TD and PV (Fst = 0.125) is near a Carbohydrate sulfotransferase, which is involved in detoxification processes. Comparing TD to RK1 and RK2 reveals multiple strong signals of selection, some which are present in both Northeastern populations, as well as many unique RK2-specific signals (Fig. 3, Supplementary Fig. 5).Fig. 3: Signals of selection over a single autosomal contig.Pairwise Fst was calculated in 1000 SNP windows over autosomal contig KB664266, comparing the Thmar Da population to the three other populations, Ratanakiri 2, Ratanakiri 1, and Preah Vihear. There is almost no indication of selection when comparing Thmar Da with Preah Vihear. There is a strongly supported signal of differentiation in both Ratanakiri 1 and Ratanakiri 2 populations at 7.5 Mbp, which is in the same location as a cluster of GSTe genes, including GSTe2, which are known to be involved in metabolic resistance to DDT and pyrethroids. The signal with the highest Fst peak here in RK2, at 6 Mbp is close to an Ecdysteroid UDP-glucosyltransferase gene, shown to confer pyrethroid insecticide resistance in other anophelines. These are a few of many selection signals identified in this study that may be associated with insecticide pressure on these An. minimus populations.Full size imageMany of the strongest signals identified in this study may be associated with insecticide pressure on these An. minimus populations. The strongest selection signals in every population comparison were close to genes that are involved in detoxification, signal transduction, and adaptations to oxidative stress, or have been functionally validated to have mutations that confer resistance to insecticides (Table 1). Some signals of interest include a strongly supported signal of selection in both RK1 and RK2 populations at 7.5 Mbp on the contig KB664266, which is in the same location as a cluster of glutathione-S-transferases, including GSTe2, which has been shown to be involved in the metabolism of DDT and pyrethroids, mutations in which mediate metabolic insecticide resistance29. The signal with the highest pairwise Fst peak on the same contig KB664266, at 6 Mbp is an RK2-specific signal and close to an Ecdysteroid UDP-glucosyltransferase gene, which has been shown to confer pyrethroid insecticide resistance in An. stephensi30.Another notable signal is between the RK1 and RK2 populations on the contig KB663610, a Peptidase S1 domain-containing protein AMIN002286, which has been shown to be involved in response to parasite pathogens in insects31. The signals of selection observed in this study are mostly distinct from the main selection signals seen in An. gambiae complex mosquitoes19, the primary vectors of Plasmodium falciparum in Africa.Insecticide resistanceWe report here variants at known insecticide resistance-associated alleles for each of the four An. minimus populations. Variants occurring at a frequency of more than 2% in at least one of the four populations are reported in the known insecticide-resistance-associated genes Ace1, Rdl, KDR, and GSTe2 (Supplementary Data 2). GSTe2 mutants are present in multiple populations, at a low rate, and there are a few individuals in Thmar Da and Preah Vihear with the Rdl resistance mutation, which is known to confer resistance to cyclodiene insecticides, despite evidence from other studies that species in this region lack this resistance mutation32. We did not investigate copy number variation, which is one mechanism by which GSTe2 confers insecticide resistance. These SNP variants indicate variation throughout these insecticide-resistance-associated genes, and though most of these populations do not currently have high rates of validated insecticide resistance-associated mutations, this underlying variation provides the potential for structural and transcriptional events resulting in greater levels of insecticide resistance in An. minimus populations. More

Dataset descriptionSample collectionOur research complies with all relevant ethical regulations following policies at the University of California, San Diego (UCSD). Animal samples that were sequenced were not collected at UCSD and are not for vertebrate animals research at UCSD following the UCSD Institutional Animal Care and Use Committee (IACUC). Samples were contributed by 34 principal investigators of the Earth Microbiome Project 500 (EMP500) Consortium and are samples from studies at their respective institutions (Supplementary Table 1). Relevant permits and ethics information for each parent study are described in the ‘Permits for sample collection’ section below. Samples were contributed as distinct sets referred to here as studies, where each study represented a single environment (for example, terrestrial plant detritus). To achieve more even coverage across microbial environments, we devised an ontology of sample types (microbial environments), the EMP Ontology (EMPO) (http://earthmicrobiome.org/protocols-and-standards/empo/)1, and selected samples to fill out EMPO categories as broadly as possible. EMPO recognizes strong gradients structuring microbial communities globally, and thus classifies microbial environments (level 4) on the basis of host association (level 1), salinity (level 2), host kingdom (if host-associated) or phase (if free-living) (level 3) (Fig. 1a). As we anticipated previously1, we have updated the number of levels as well as states therein for EMPO (Fig. 1b) on the basis of an important additional salinity gradient observed among host-associated samples when considering the previously unreported shotgun metagenomic and metabolomic data generated here (Fig. 3c,d). We note that although we were able to acquire samples for all EMPO categories, some categories are represented by a single study.Samples were collected following the Earth Microbiome Project sample submission guide50. Briefly, samples were collected fresh, split into 10 aliquots and then frozen, or alternatively collected and frozen, and subsequently split into 10 aliquots with minimal perturbation. Aliquot size was sufficient to yield 10–100 ng genomic DNA (approximately 107–108 cells). To leave samples amenable to chemical characterization (metabolomics), buffers or solutions for sample preservation (for example, RNAlater) were avoided. Ethanol (50–95%) was allowed as it is compatible with LC–MS/MS although it should also be avoided if possible.Sampling guidance was tailored for four general sample types: bulk unaltered (for example, soil, sediment, faeces), bulk fractionated (for example, sponges, corals, turbid water), swabs (for example, biofilms) and filters. Bulk unaltered samples were split fresh (or frozen), sampled into 10 pre-labelled 2 ml screw-cap bead beater tubes (Sarstedt, 72.694.005 or similar), ideally with at least 200 mg biomass, and flash frozen in liquid nitrogen (if possible). Bulk fractionated samples were fractionated as appropriate for the sample type, split into 10 pre-labelled 2 ml screw-cap bead beater tubes, ideally with at least 200 mg biomass, and flash frozen in liquid nitrogen (if possible). Swabs were collected as 10 replicate swabs using 5 BD SWUBE dual cotton swabs with wooden stick and screw cap (281130). Filters were collected as 10 replicate filters (47 mm diameter, 0.2 um pore size, polyethersulfone (preferred) or hydrophilic PTFE filters), placed in pre-labelled 2 ml screw-cap bead beater tubes, and flash frozen in liquid nitrogen (if possible). All sample types were stored at –80 °C if possible, otherwise –20 °C.To track the provenance of sample aliquots, we employed a QR coding scheme. Labels were affixed to aliquot tubes before shipping when possible. QR codes had the format ‘name.99.s003.a05’, where ‘name’ is the PI name, ‘99’ is the study ID, ‘s003’ is the sample number and ‘a05’ is the aliquot number. QR codes (version 2, 25 pixels × 25 pixels) were printed on 1.125’ × 0.75’ rectangular and 0.437’ circular cap Cryogenic Direct Thermal labels (GA International, DFP-70) using a Zebra model GK420d printer and ZebraDesigner Pro 3 software for Windows. After receipt but before aliquots were stored in freezers, QR codes were scanned into a sample inventory spreadsheet using a QR scanner.Sample metadataEnvironmental metadata were collected for all samples on the basis of the EMP Metadata Guide, which combines guidance from the Genomics Standards Consortium MIxS (Minimum Information about any Sequence) standard74 and the Qiita Database (https://qiita.ucsd.edu)51. The metadata guide provides templates and instructions for each MIxS environmental package (that is, sample type). Relevant information describing each PI submission, or study, was organized into a separate study metadata file (Supplementary Table 1).MetabolomicsLC–MS/MS sample extraction and preparationTo profile metabolites among all samples, we used LC–MS/MS, a versatile method that detects tens of thousands of metabolites in biological samples. All solvents and reactants used were LC–MS grade. To maximize the biomass extracted from each sample, the samples were prepared depending on their sampling method (for example, bulk, swabs, filter and controls). The bulk samples were transferred into a microcentrifuge tube (polypropylene, PP) and dissolved in 7:3 MeOH:H2O using a volume varying from 600 µl to 1.5 ml, depending on the amounts of sample available, and homogenized in a tissue lyser (QIAGEN) at 25 Hz for 5 min. Then, the tubes were centrifuged at 2,000 × g for 15 min, and the supernatant was collected in a 96-well plate (PP). For swabs, the swabs were transferred into a 96-well plate (PP) and dissolved in 1.0 ml of 9:1 ethanol:H2O. The prepared plates were sonicated for 30 min, and after 12 h at 4 °C, the swabs were removed from the wells. The filter samples were dissolved in 1.5 ml of 7:3 MeOH:H2O in microcentrifuge tubes (PP) and sonicated for 30 min. After 12 h at 4 °C, the filters were removed from the tubes. The tubes were centrifuged at 2,000 × g for 15 min, and the supernatants were transferred to 96-well plates (PP). The process control samples (bags, filters and tubes) were prepared by adding 3.0 ml of 2:8 MeOH:H2O and recovering 1.5 ml after 2 min. After the extraction process, all sample plates were dried with a vacuum concentrator and subjected to solid phase extraction (SPE). SPE was used to remove salts that could reduce ionization efficiency during mass spectrometry analysis, as well as the most polar and non-polar compounds (for example, waxes) that cannot be analysed efficiently by reversed-phase chromatography. The protocol was as follows: the samples (in plates) were dissolved in 300 µl of 7:3 MeOH:H2O and put in an ultrasound bath for 20 min. SPE was performed with SPE plates (Oasis HLB, hydrophilic-lipophilic-balance, 30 mg with particle sizes of 30 µm). The SPE beds were activated by priming them with 100% MeOH, and equilibrated with 100% H2O. The samples were loaded on the SPE beds, and 100% H2O was used as wash solvent (600 µl). The eluted washing solution was discarded, as it contains salts and very polar metabolites that subsequent metabolomics analysis is not designed for. The sample elution was carried out sequentially with 7:3 MeOH:H2O (600 µl) and 100% MeOH (600 µl). The obtained plates were dried with a vacuum concentrator. For mass spectrometry analysis, the samples were resuspended in 130 µl of 7:3 MeOH:H2O containing 0.2 µM of amitriptyline as an internal standard. The plates were centrifuged at 30 × g for 15 min at 4 °C. Samples (100 µl) were transferred into new 96-well plates (PP) for mass spectrometry analysis.LC–MS/MS sample analysisThe extracted samples were analysed by ultra-high performance liquid chromatography (UHPLC, Vanquish, Thermo Fisher) coupled to a quadrupole-Orbitrap mass spectrometer (Q Exactive, Thermo Fisher) operated in data-dependent acquisition mode (LC–MS/MS in DDA mode). Chromatographic separation was performed using a Kinetex C18 1.7 µm (Phenomenex), 100 Å pore size, 2.1 mm (internal diameter) × 50 mm (length) column with a C18 guard cartridge (Phenomenex). The column was maintained at 40 °C. The mobile phase was composed of a mixture of (A) water with 0.1% formic acid (v/v) and (B) acetonitrile with 0.1% formic acid. Chromatographic elution method was set as follows: 0.00–1.00 min, isocratic 5% B; 1.00–9.00 min, gradient from 5% to 100% B; 9.00–11.00 min, isocratic 100% B; followed by equilibration 11.00–11.50 min, gradient from 100% to 5% B; 11.50–12.50 min, isocratic 5% B. The flow rate was set to 0.5 ml min−1.The UHPLC was interfaced to the orbitrap using a heated electrospray ionization source with the following parameters: ionization mode, positive; spray voltage, +3,496.2 V; heater temperature, 363.90 °C; capillary temperature, 377.50 °C; S-lens RF, 60 arbitrary units (a.u.); sheath gas flow rate, 60.19 a.u.; and auxiliary gas flow rate, 20.00 a.u. The MS1 scans were acquired at a resolution (at m/z 200) of 35,000 in the m/z 100–1500 range, and the fragmentation spectra (MS2) scans at a resolution of 17,500 from 0 to 12.5 min. The automatic gain control target and maximum injection time were set at 1.0 × 106 and 160 ms for MS1 scans, and set at 5.0 × 105 and 220 ms for MS2 scans, respectively. Up to three MS2 scans in data-dependent mode (Top 3) were acquired for the most abundant ions per MS1 scans using the apex trigger mode (4–15 s), dynamic exclusion (11 s) and automatic isotope exclusion. The starting value for MS2 was m/z 50. Higher-energy collision induced dissociation (HCD) was performed with a normalized collision energy of 20, 30 and 40 eV in stepped mode. The major background ions originating from the SPE were excluded manually from the MS2 acquisition. Analyses were randomized within plate and blank samples analysed every 20 injections. A quality control mix sample assembled from 20 random samples across the sample types was injected at the beginning, the middle and the end of each plate sequence. The chromatographic shift observed throughout the batch was estimated as less than 2 s, and the relative standard deviation of ion intensity was 15% per replicate.LC–MS/MS data processingThe mass spectrometry data were centroided and converted from the proprietary format (.raw) to the m/z extensible markup language format (.mzML) using ProteoWizard (ver. 3.0.19, MSConvert tool)75. The mzML files were then processed with MZmine 2 toolbox76 using the ion-identity networking modules77 that allow advanced detection for adduct/isotopologue annotations. The MZmine processing was performed on Ubuntu 18.04 LTS 64-bits workstation (Intel Xeon E5-2637, 3.5 GHz, 8 cores, 64 Gb of RAM) and took ~3 d. The MZmine project, the MZmine batch file (.XML format) and results files (.MGF and .CSV) are available in the MassIVE dataset MSV000083475. The MZmine batch file contains all the parameters used during the processing. In brief, feature detection and deconvolution was performed with the ADAP chromatogram builder78 and local minimum search algorithm. The isotopologues were regrouped and the features (peaks) were aligned across samples. The aligned peak list was gap filled and only peaks with an associated fragmentation spectrum and occurring in a minimum of three files were conserved. Peak shape correlation analysis grouped peaks originating from the same molecule and annotated adduct/isotopologue with ion-identity networking77. Finally, the feature quantification table results (.CSV) and spectral information (.MGF) were exported with the GNPS module for feature-based molecular networking analysis on GNPS79 and with SIRIUS export modules.LC–MS/MS data annotationThe results files of MZmine (.MGF and .CSV files) were uploaded to GNPS (http://gnps.ucsd.edu)52 and analysed with the feature-based molecular networking workflow79. Spectral library matching was performed against public fragmentation spectra (MS2) spectral libraries on GNPS and the NIST17 library.For the additional annotation of small peptides, we used the DEREPLICATOR tools available on GNPS80,81. We then used SIRIUS82 (v. 4.4.25, headless, Linux) to systematically annotate the MS2 spectra. Molecular formulae were computed with the SIRIUS module by matching the experimental and predicted isotopic patterns83, and from fragmentation trees analysis84 of MS2. Molecular formula prediction was refined with the ZODIAC module using Gibbs sampling85 on the fragmentation spectra (chimeric spectra or those with poor fragmentation were excluded). In silico structure annotation using structures from biodatabase was done with CSI:FingerID86. Systematic class annotations were obtained with CANOPUS41 and used the NPClassifier ontology87.The parameters for SIRIUS tools were set as follows, for SIRIUS: molecular formula candidates retained, 80; molecular formula database, ALL; maximum precursor ion m/z computed, 750; profile, orbitrap; m/z maximum deviation, 10 ppm; ions annotated with MZmine were prioritized and other ions were considered (that is, [M+H3N+H]+, [M+H]+, [M+K]+, [M+Na]+, [M+H-H2O]+, [M+H-H4O2]+, [M+NH4]+); for ZODIAC: the features were split into 10 random subsets for lower computational burden and computed separately with the following parameters: threshold filter, 0.9; minimum local connections, 0; for CSI:FingerID: m/z maximum deviation, 10 ppm; and biological database, BIO.To establish putative microbially related secondary metabolites, we collected annotations from spectral library matching and the DEREPLICATOR+ tools and queried them against the largest microbial metabolite reference databases (Natural Products Atlas88 and MIBiG89). Molecular networking79 was then used to propagate the annotation of microbially related secondary metabolites throughout all molecular families (that is, the network component).LC–MS/MS data analysisWe combined the annotation results from the different tools described above to create a comprehensive metadata file describing each metabolite feature observed. Using that information, we generated a feature-table including only secondary metabolite features determined to be microbially related. We then excluded very low-intensity features introduced to certain samples during the gap-filling step described above. These features were identified on the basis of presence in negative controls that were universal to all sample types (that is, bulk, filter and swab) and by their relatively low per-sample intensity values. Finally, we excluded features present in positive controls for sampling devices specific to each sample type (that is, bulk, filter or swab). The final feature-table included 618 samples and 6,588 putative microbially related secondary metabolite features that were used for subsequent analysis.We used QIIME 2’s90 (v2020.6) ‘diversity’ plugin to quantify alpha-diversity (that is, feature richness) for each sample and ‘deicode’91 to quantify beta-diversity (that is, robust Aitchison distances, which are robust to both sparsity and compositionality in the data) between each pair of samples. We parameterized our robust Aitchison principal components analysis (RPCA)91 to exclude samples with fewer than 500 features and features present in fewer than 10% of samples. We used the ‘taxa’ plugin to quantify the relative abundance of microbially related secondary metabolite pathways and superclasses (that is, on the basis of NPClassifier) within each environment (that is, for each level of EMPO 4), and ‘songbird’ v1.0.492 to identify sets of microbially related secondary metabolites whose abundances were associated with certain environments. We parameterized our ‘songbird’ model as follows: epochs, 1,000,000; differential prior, 0.5; learning rate, 1.0 × 10−5; summary interval, 2; batch size, 400; minimum sample count, 0; and training on 80% of samples at each level of EMPO 4 using ‘Animal distal gut (non-saline)’ as the reference environment. Environments with fewer than 10 samples were excluded to optimize model training (that is, ‘Animal corpus (non-saline)’, ‘Animal proximal gut (non-saline)’, ‘Surface (saline)’). The output from ‘songbird’ includes a rank value for each metabolite in every environment, which represents the log fold change for a given metabolite in a given environment92. We compared log fold changes for each metabolite from this run to those from (1) a replicate run using the same reference environment and (2) a run using a distinct reference environment: ‘Water (saline)’. We found strong Spearman correlations in both cases (Supplementary Table 8), and therefore focused on results from the original run using ‘Animal distal gut (non-saline)’ as the reference environment, as it has previously been shown to be relatively unique among other habitats. In addition to summarizing the top 10 metabolites for each environment (Supplementary Table 3), we used the log fold change values in our multi-omics analyses described below.We used the RPCA biplot and QIIME 2’s90 EMPeror93 to visualize differences in composition among samples, as well as the association with samples of the 25 most influential microbially related secondary metabolite features (that is, those with the largest magnitude across the first three principal component loadings). We tested for significant differences in metabolite composition across all levels of EMPO using PERMANOVA implemented with QIIME 2’s ‘diversity’ plugin90 and using our robust Aitchison distance matrix as input. In parallel, we used the differential abundance results from ‘songbird’ described above to identify specific microbially related secondary metabolite pathways and superclasses that varied strongly across environments. We then went back to our metabolite feature-table to visualize differences in the relative abundances of those pathways and superclasses within each environment by first selecting features and calculating log-ratios using ‘qurro’94, and then plotting using the ‘ggplot2’ package95 in R96 v4.0.0. We tested for significant differences in relative abundances across environments using Kruskal–Wallis tests implemented with the base ‘stats’ package in R96.GC–MS sample extraction and preparationTo profile volatile small molecules among all samples in addition to what was captured with LC–MS/MS, we used gas chromatography coupled with mass spectrometry (GC–MS). All solvents and reactants were GC–MS grade. Two protocols were used for sample extraction, one for the 105 soil samples and a second for the 356 faecal and sediment samples that were treated as biosafety level 2. The 105 soil samples were received at the Pacific Northwest National Laboratory and processed as follows. Each soil sample (1 g) was weighed into microcentrifuge tubes (Biopur Safe-Lock, 2.0 ml, Eppendorf). H2O (1 ml) and one scoop (~0.5 g) of a 1:1 (v/v) mixture of garnet (0.15 mm, Omni International) and stainless steel (0.9–2.0 mm blend, Next Advance) beads and one 3 mm stainless steel bead (Qiagen) were added to each tube. Samples were homogenized in a tissue lyser (Qiagen) for 3 min at 30 Hz and transferred into 15 ml polypropylene tubes (Olympus, Genesee Scientific). Ice-cold water (1 ml) was used to rinse the smaller tube and combined into the 15 ml tube. Chloroform:methanol (10 ml, 2:1 v/v) was added and samples were rotated at 4 °C for 10 min, followed by cooling at −70 °C for 10 min and centrifuging at 150 × g for 10 min to separate phases. The top and bottom layers were combined into 40 ml glass vials and dried using a vacuum concentrator. Chloroform:methanol (1 ml, 2:1) was added to each large glass vial and the sample was transferred into 1.5 ml tubes and centrifuged at 1,300 × g. The supernatant was transferred into glass vials and dried for derivatization.The remaining 356 samples received from UCSD that included faecal and sediment samples were processed as follows: 100 µl of each sample was transferred to a 2 ml microcentrifuge tube using a scoop (MSP01, Next Advance). The final volume of the sample was brought to 1.5 ml, ensuring that the solvent ratio is 3:8:4 H2O:CHCl3:MeOH by adding the appropriate volumes of H2O, MeOH and CHCl3. After transfer, one 3 mm stainless steel bead (QIAGEN), 400 µl methanol and 300 µl H2O were added to each tube and the samples were vortexed for 30 s. Then, 800 µl chloroform was added and samples were vortexed for 30 s. After centrifuging at 150 × g for 10 min to separate phases, the top and bottom layers were combined in a vial and dried for derivatization.The samples were derivatized for GC–MS analysis as follows: 20 µl of a methoxyamine solution in pyridine (30 mg ml−1) was added to the sample vial and vortexed for 30 s. A bath sonicator was used to ensure that the sample was completely dissolved. Samples were incubated at 37 °C for 1.5 h while shaking at 1,000 r.p.m. N-methyl-N-trimethylsilyltrifluoroacetamide (80 µl) and 1% trimethylchlorosilane solution was added and samples were vortexed for 10 s, followed by incubation at 37 °C for 30 min, with 1,000 r.p.m. shaking. The samples were then transferred into a vial with an insert.An Agilent 7890A gas chromatograph coupled with a single quadrupole 5975C mass spectrometer (Agilent) and an HP-5MS column (30 m × 0.25 mm × 0.25 μm; Agilent) was used for untargeted analysis. Samples (1 μl) were injected in splitless mode, and the helium gas flow rate was determined by the Agilent Retention Time Locking function on the basis of analysis of deuterated myristic acid (Agilent). The injection port temperature was held at 250 °C throughout the analysis. The GC oven was held at 60 °C for 1 min after injection, and the temperature was then increased to 325 °C at a rate of 10 °C min−1, followed by a 10 min hold at 325 °C. Data were collected over the mass range of m/z 50–600. A mixture of FAMEs (C8–C28) was analysed each day with the samples for retention index alignment purposes during subsequent data analysis.GC–MS data processing and annotationThe data were converted from vendor’s format to the .mzML format and processed using GNPS GC–MS data analysis workflow (https://gnps.ucsd.edu)97. The compounds were identified by matching experimental spectra to the public libraries available at GNPS, as well as NIST 17 and Wiley libraries. The data are publicly available at the MassIVE depository (https://massive.ucsd.edu); dataset ID: MSV000083743. The GNPS deconvolution is available in GNPS (https://gnps.ucsd.edu/ProteoSAFe/status.jsp?task=d5c5135a59eb48779216615e8d5cb3ac), as is the library search (https://gnps.ucsd.edu/ProteoSAFe/status.jsp?task=59b20fc8381f4ee6b79d35034de81d86).GC–MS data analysisFor multi-omics analyses including GC–MS data, we first removed noisy (that is, suspected background contaminants and artifacts) features by excluding those with balance scores 1.5–2 kb DNA fragments’ (Oxford Nanopore Technologies). The resulting product consists of uniquely tagged rRNA operon amplicons. The uniquely tagged rRNA operons were amplified in a second PCR, where the reaction (100 µl) contained 2 U Platinum SuperFi DNA Polymerase High Fidelity (Thermo Fisher) and a final concentration of 1X SuperFi buffer, 0.2 mM of each dNTP, and 500 nM of each forward and reverse synthetic primer targeting the tailed primers from above. The PCR cycling parameters consisted of an initial denaturation (3 min at 95 °C) and then 25–35 cycles of denaturation (15 s at 95 °C), annealing (30 s at 60 °C) and extension (6 min at 72 °C), followed by final extension (5 min at 72 °C). The PCR product was purified using the custom bead purification protocol above. Batches of 25 amplicon libraries were barcoded and sent for PacBio Sequel II library preparation and sequencing (Sequel II SMRT Cell 8M and 30 h collection time) at the DNA Sequencing Center at Brigham Young University. Circular consensus sequencing (CCS) reads were generated using CCS v.3.4.1 (https://github.com/PacificBiosciences/ccs) using default settings. UMI consensus sequences were generated using the longread_umi pipeline (https://github.com/SorenKarst/longread_umi) with the following command: longread_umi pacbio_pipeline -d ccs_reads.fq -o out_dir -m 3500 -M 6000 -s 60 -e 60 -f CAAGCAGAAGACGGCATACGAGAT -F AGRGTTYGATYMTGGCTCAG -r AATGATACGGCGACCACCGAGATC -R CGACATCGAGGTGCCAAAC -U ‘0.75;1.5;2;0’ -c 2.Amplicon data analysisFor multi-omics analyses including amplicon sequence data, we processed each dataset for comparison of beta-diversity. For all amplicon data except that for bacterial full-length rRNA amplicons, raw sequence data were converted from bcl to fastq, and then multiplexed files for each sequencing run uploaded as separate preparations to Qiita (study: 13114).For each 16S sequencing run, in Qiita, data were demultiplexed, trimmed to 150 bp and denoised using Deblur122 to generate a feature-table of sub-operational taxonomic units (sOTUs) per sample, using default parameters. We then exported feature-tables and denoised sequences from each sequencing run, used QIIME 2’s ‘feature-table’ plugin to merge feature-tables and denoised reads across sequencing runs, and placed all denoised reads into the GreenGenes 13_8 phylogeny123 via fragment insertion using QIIME 2’s90 SATé-Enabled Phylogenetic Placement (SEPP)124 plugin to produce a phylogeny for diversity analyses. To allow for phylogenetically informed diversity analyses, reads not placed during SEPP (that is, 513 sOTUs, 0.1% of all sOTUs) were removed from the merged feature-table. We then used QIIME 2’s ‘feature-table’ plugin to exclude singleton sOTUs and rarefy the data to 5,000 reads per sample. Rarefaction depths for all amplicon analyses were chosen to best normalize sampling effort per sample while maintaining ≥75% of samples representative of Earth’s environments, and also to maintain consistency with the analyses from EMP release 1. We then used QIIME 2’s90 ‘diversity’ plugin to estimate alpha-diversity (that is, sOTU richness) and beta-diversity (that is, unweighted UniFrac distances). The final feature-table for 16S beta-diversity analysis included 681 samples and 93,260 features. We performed a comparative analysis of the data including and excluding the reads not placed during SEPP, and note that both alpha-diversity (that is, sOTU richness) and beta-diversity (that is, sample–sample RPCA distances) were highly correlated between datasets (Spearman r = 1.0) (Supplementary Fig. 5). We thus proceeded with the SEPP-filtered dataset and used phylogenetically informed diversity metrics where applicable.For 18S data, we used QIIME 2’s90 ‘demux’ plugin’s ‘emp-paired’ method125,126 to first demultiplex each sequencing run, and then the ‘cutadapt’ plugin’s127 ‘trim-paired’ method to trim sequencing primers from reads. We then exported trimmed reads, concatenated R1 and R2 read files per sample, and denoised reads using Deblur’s122,128 ‘workflow’ with default settings, trimming reads to 90 bp, and taking the ‘all.biom’ and ‘all.seqs’ output, for each sequencing run. We then used QIIME 2’s ‘feature-table’ plugin to merge feature-tables and denoised sequences across sequencing runs, and then the ‘feature-classifier’ plugin’s ‘classify-sklearn’ method to classify taxonomy for each sOTU via pre-fitted machine-learning classifiers129 and the SILVA 138 reference database130. We then used QIIME 2’s90 ‘feature-table’ plugin to exclude reads assigned to bacteria and archaea, singleton sOTUs and samples with a total frequency of More

Keenan, T. F. & Williams, C. A. The terrestrial carbon sink. Annu. Rev. Environ. Resour. 43, 219–243 (2018).Article 

Google Scholar 
Terrer, C. et al. A trade-off between plant and soil carbon storage under elevated CO2. Nature 591, 599–603 (2021).Article 

Google Scholar 
Walker, A. P. et al. Integrating the evidence for a terrestrial carbon sink caused by increasing atmospheric CO2. N. Phytol. 229, 2413–2445 (2021).Article 

Google Scholar 
Fossum, C. et al. Belowground allocation and dynamics of recently fixed plant carbon in a California annual grassland. Soil Biol. Biochem. 165, 108519 (2022).Article 

Google Scholar 
Rasse, D. P., Rumpel, C. & Dignac, M.-F. Is soil carbon mostly root carbon? Mechanisms for a specific stabilisation. Plant Soil 269, 341–356 (2005).Article 

Google Scholar 
Sokol, N. W., Kuebbing, Sara, E., Karlsen-Ayala, E. & Bradford, M. A. Evidence for the primacy of living root inputs, not root or shoot litter, in forming soil organic carbon. N. Phytol. 221, 233–246 (2019).Article 

Google Scholar 
Calvo, O. C., Franzaring, J., Schmid, I. & Fangmeier, A. Root exudation of carbohydrates and cations from barley in response to drought and elevated CO2. Plant Soil 438, 127–142 (2019).Article 

Google Scholar 
Fransson, P. M. A. & Johansson, E. M. Elevated CO2 and nitrogen influence exudation of soluble organic compounds by ectomycorrhizal root systems. FEMS Microbiol. Ecol. 71, 186–196 (2009).Article 

Google Scholar 
Johansson, E. M., Fransson, P. M. A., Finlay, R. D. & van Hees, P. A. W. Quantitative analysis of soluble exudates produced by ectomycorrhizal roots as a response to ambient and elevated CO2. Soil Biol. Biochem. 41, 1111–1116 (2009).Article 

Google Scholar 
Phillips, R. P., Finzi, A. C. & Bernhardt, E. S. Enhanced root exudation induces microbial feedbacks to N cycling in a pine forest under long-term CO2 fumigation. Ecol. Lett. 14, 187–194 (2011).Article 

Google Scholar 
Jilling, A., Keiluweit, M., Gutknecht, J. L. M. & Grandy, A. S. Priming mechanisms providing plants and microbes access to mineral-associated organic matter. Soil Biol. Biochem. 158, 108265 (2021).Article 

Google Scholar 
Cotrufo, M. F., Wallenstein, M. D., Boot, C. M., Denef, K. & Paul, E. The Microbial Efficiency-Matrix Stabilization (MEMS) framework integrates plant litter decomposition with soil organic matter stabilization: do labile plant inputs form stable soil organic matter? Glob. Change Biol. 19, 988–995 (2013).Article 

Google Scholar 
Sokol, N. W., Sanderman, J. & Bradford, M. A. Pathways of mineral-associated soil organic matter formation: integrating the role of plant carbon source, chemistry, and point of entry. Glob. Change Biol. 25, 12–24 (2019).Article 

Google Scholar 
Bradford, M. A., Keiser, A. D., Davies, C. A., Mersmann, C. A. & Strickland, M. S. Empirical evidence that soil carbon formation from plant inputs is positively related to microbial growth. Biogeochemistry 113, 271–281 (2013).Article 

Google Scholar 
Keiluweit, M. et al. Mineral protection of soil carbon counteracted by root exudates. Nat. Clim. Change 5, 588–595 (2015).Article 

Google Scholar 
Kuzyakov, Y., Friedel, J. K. & Stahr, K. Review of mechanisms and quantification of priming effects. Soil Biol. Biochem. 32, 1485–1498 (2000).Article 

Google Scholar 
Jones, D. L., Dennis, P. G., Owen, A. G. & van Hees, P. A. W. Organic acid behavior in soils—misconceptions and knowledge gaps. Plant Soil 248, 31–41 (2003).Article 

Google Scholar 
Cleveland, C. C. & Liptzin, D. C:N:P stoichiometry in soil: is there a “Redfield ratio” for the microbial biomass? Biogeochemistry 85, 235–252 (2007).Article 

Google Scholar 
Meier, I. C., Finzi, A. C. & Phillips, R. P. Root exudates increase N availability by stimulating microbial turnover of fast-cycling N pools. Soil Biol. Biochem. 106, 119–128 (2017).Article 

Google Scholar 
Canarini, A., Kaiser, C., Merchant, A., Richter, A. & Wanek, W. Root exudation of primary metabolites: mechanisms and their roles in plant responses to environmental stimuli. Front. Plant Sci. 10, 157 (2019).Article 

Google Scholar 
Koo, B.-J., Adriano, D. C., Bolan, N. S. & Barton, C. D. in Encyclopedia of Soils in the Environment (ed. Hillel, D.) 421–428 (Elsevier, 2005); https://doi.org/10.1016/B0-12-348530-4/00461-6Oldfield, E. E., Crowther, T. W. & Bradford, M. A. Substrate identity and amount overwhelm temperature effects on soil carbon formation. Soil Biol. Biochem. 124, 218–226 (2018).Article 

Google Scholar 
Mason-Jones, K., Schmücker, N. & Kuzyakov, Y. Contrasting effects of organic and mineral nitrogen challenge the N-mining hypothesis for soil organic matter priming. Soil Biol. Biochem. 124, 38–46 (2018).Article 

Google Scholar 
Sokol, N. W. & Bradford, M. A. Microbial formation of stable soil carbon is more efficient from belowground than aboveground input. Nat. Geosci. 12, 46–53 (2019).Article 

Google Scholar 
Drake, J. E. et al. Stoichiometry constrains microbial response to root exudation—insights from a model and a field experiment in a temperate forest. Biogeosciences 10, 821–838 (2013).Article 

Google Scholar 
Falchini, L., Naumova, N., Kuikman, P. J., Bloem, J. & Nannipieri, P. CO2 evolution and denaturing gradient gel electrophoresis profiles of bacterial communities in soil following addition of low molecular weight substrates to simulate root exudation. Soil Biol. Biochem. 35, 775–782 (2003).Article 

Google Scholar 
Rasmussen, C., Southard, R. J. & Horwath, W. R. Soil mineralogy affects conifer forest soil carbon source utilization and microbial priming. Soil Sci. Soc. Am. J. 71, 1141–1150 (2007).Article 

Google Scholar 
Frey, S. D., Lee, J., Melillo, J. M. & Six, J. The temperature response of soil microbial efficiency and its feedback to climate. Nat. Clim. Change 3, 395–398 (2013).Article 

Google Scholar 
Angst, G., Mueller, K. E., Nierop, K. G. J. & Simpson, M. J. Plant- or microbial-derived? A review on the molecular composition of stabilized soil organic matter. Soil Biol. Biochem. 156, 108189 (2021).Article 

Google Scholar 
Craig, M. E. et al. Fast-decaying plant litter enhances soil carbon in temperate forests but not through microbial physiological traits. Nat. Commun. 13, 1229 (2022).Article 

Google Scholar 
Blagodatsky, S., Blagodatskaya, E., Yuyukina, T. & Kuzyakov, Y. Model of apparent and real priming effects: linking microbial activity with soil organic matter decomposition. Soil Biol. Biochem. 42, 1275–1283 (2010).Article 

Google Scholar 
Hill, P. W., Farrar, J. F. & Jones, D. L. Decoupling of microbial glucose uptake and mineralization in soil. Soil Biol. Biochem. 40, 616–624 (2008).Article 

Google Scholar 
Asmar, F., Eiland, F. & Nielsen, N. E. Interrelationship between extracellular enzyme activity, ATP content, total counts of bacteria and CO2 evolution. Biol. Fertil. Soils 14, 288–292 (1992).Article 

Google Scholar 
Fontaine, S., Mariotti, A. & Abbadie, L. The priming effect of organic matter: a question of microbial competition? Soil Biol. Biochem. 35, 837–843 (2003).Article 

Google Scholar 
McFarlane, K. J. et al. Comparison of soil organic matter dynamics at five temperate deciduous forests with physical fractionation and radiocarbon measurements. Biogeochemistry 112, 457–476 (2013).Article 

Google Scholar 
Post, W. M., Emanuel, W. R., Zinke, P. J. & Stangenberger, A. G. Soil carbon pools and world life zones. Nature 298, 156–159 (1982).Article 

Google Scholar 
Smith, W. H. Character and significance of forest tree root exudates. Ecology 57, 324–331 (1976).Article 

Google Scholar 
Dong, J. et al. Impacts of elevated CO2 on plant resistance to nutrient deficiency and toxic ions via root exudates: a review. Sci. Total Environ. 754, 142434 (2021).Article 

Google Scholar 
White, M. A., Running, S. W. & Thornton, P. E. The impact of growing-season length variability on carbon assimilation and evapotranspiration over 88 years in the eastern US deciduous forest. Int. J. Biometeorol. 42, 139–145 (1999).Article 

Google Scholar 
Giasson, M.-A. et al. Soil respiration in a northeastern US temperate forest: a 22-year synthesis. Ecosphere 4, 140 (2013).Article 

Google Scholar 
Mrak, T. et al. Elevated ozone prevents acquisition of available nitrogen due to smaller root surface area in poplar. Plant Soil 450, 585–599 (2020).Article 

Google Scholar 
Cotrufo, M. F., Ranalli, M. G., Haddix, M. L., Six, J. & Lugato, E. Soil carbon storage informed by particulate and mineral-associated organic matter. Nat. Geosci. 12, 989–994 (2019).Article 

Google Scholar 
Brookes, P. C., Landman, A., Pruden, G. & Jenkinson, D. S. Chloroform fumigation and the release of soil nitrogen: a rapid direct extraction method to measure microbial biomass nitrogen in soil. Soil Biol. Biochem. 17, 837–842 (1985).Article 

Google Scholar 
Haney, R. L., Franzluebbers, A. J., Hons, F. M. & Zuberer, D. A. Soil C extracted with water or K2SO4: pH effect on determination of microbial biomass. Can. J. Soil Sci. 79, 529–533 (1999).Article 

Google Scholar 
Ahmed, M. J. & Hossan, J. Spectrophotometric determination of aluminium by morin. Talanta 42, 1135–1142 (1995).Article 

Google Scholar 
Viollier, E., Inglett, P. W., Hunter, K., Roychoudhury, A. N. & Van Cappellen, P. The ferrozine method revisited: Fe(II)/Fe(III) determination in natural waters. Appl. Geochem. 15, 785–790 (2000).Article 

Google Scholar  More

Curtis, P. G., Slay, C. M., Harris, N. L., Tyukavina, A. & Hansen, M. C. Classifying drivers of global forest loss. Science 361, 1108–1111 (2018).CAS 
PubMed 

Google Scholar 
Laurance, W. F., Sayer, J. & Cassman, K. G. Agricultural expansion and its impacts on tropical nature. Trends Ecol. Evol. 29, 107–116 (2014).PubMed 

Google Scholar 
Pendrill, F. et al. Agricultural and forestry trade drives large share of tropical deforestation emissions. Glob. Environ. Change 56, 1–10 (2019).
Google Scholar 
Haupt, F., Bakhtary, H., Schulte, I., Galt, H. & Streck, C. Progress on Corporate Commitments and their Implementation (Tropical Forest Alliance, 2018); https://www.tropicalforestalliance.org/assets/Uploads/Progress-on-Corporate-Commitments-and-their-Implementation.pdfAustin, K. G. et al. Mapping and monitoring zero-deforestation commitments. Bioscience 71, 1079–1090 (2021).PubMed 
PubMed Central 

Google Scholar 
Leijten, F. C., Sim, S., King, H. & Verburg, P. H. Which forests could be protected by corporate zero deforestation commitments? A spatial assessment. Environ. Res. Lett. 15, 064021 (2020).
Google Scholar 
Garrett, R. D. et al. Criteria for effective zero-deforestation commitments. Glob. Environ. Change https://doi.org/10.1016/j.gloenvcha.2018.11.003 (2019).Lehmann, C. E. R. & Parr, C. L. Tropical grassy biomes: linking ecology, human use and conservation. Phil. Trans. R. Soc. B https://doi.org/10.1098/rstb.2016.0329 (2016).Miles, L. et al. A global overview of the conservation status of tropical dry forests. J. Biogeogr. 33, 491–505 (2006).
Google Scholar 
Gibbs, H. K. et al. Brazil’s soy moratorium. Science https://doi.org/10.1126/science.aaa0181 (2015).Jopke, P. & Schoneveld, G. C. Corporate Commitments to Zero Deforestation: An Evaluation of Externality Problems and Implementation Gaps (CIFOR, 2018); https://doi.org/10.17528/cifor/006827Parr, C. L., Lehmann, C. E. R., Bond, W. J., Hoffmann, W. A. & Andersen, A. N. Tropical grassy biomes: misunderstood, neglected, and under threat. Trends Ecol. Evol. https://doi.org/10.1016/j.tree.2014.02.004 (2014).Ratnam, J. et al. When is a ‘forest’ a savanna, and why does it matter? Glob. Ecol. Biogeogr. https://doi.org/10.1111/j.1466-8238.2010.00634.x (2011).Sanchez-Azofeifa, G. A. et al. Research priorities for neotropical dry forests. Biotropica 37, 477–485 (2005).
Google Scholar 
Vijay, V., Pimm, S. L., Jenkins, C. N. & Smith, S. J. The impacts of oil palm on recent deforestation and biodiversity loss. PLoS ONE 11, e0159668 (2016).PubMed 
PubMed Central 

Google Scholar 
Principles & Criteria for the Production of Sustainable Palm Oil (RSPO, 2018).Rosoman, G. et al. (eds) The HCS Approach Toolkit (HCS Approach Steering Group, 2017).Brown, E. & Senior, M. J. M. (eds) Common Guidance for the Identification of High Conservation Values (HCV Resource Network, 2017).Furumo, P. R. & Aide, T. M. Characterizing commercial oil palm expansion in Latin America: land use change and trade. Environ. Res. Lett. 12, 024008 (2017).
Google Scholar 
Dinerstein, E. et al. An ecoregion-based approach to protecting half the terrestrial realm. BioScience https://doi.org/10.1093/biosci/bix014 (2017).Descals, A. et al. High-resolution global map of smallholder and industrial closed-canopy oil palm plantations. Earth Syst. Sci. Data 13, 1211–1231 (2021).
Google Scholar 
Woittiez, L. S., van Wijk, M. T., Slingerland, M., van Noordwijk, M. & Giller, K. E. Yield gaps in oil palm: a quantitative review of contributing factors. Eur. J. Agron. 83, 57–77 (2017).
Google Scholar 
Kuepper, B., Drost, S. & Piotrowski, M. Latin American Palm Oil Linked to Social Risks, Local Deforestation (Chain Reaction Research, 2021); https://chainreactionresearch.com/wp-content/uploads/2021/12/Latin-American-Palm-Oil-Linked-to-Social-Issues-Local-Deforestation-1.pdfHoyle, D. et al. RSPO New Planting Procedures: Summary Report of ESIA, HCV Assessments and Management Plan (Terea, Proforest and Olam Palm Gabon, 2017).Universal Mill List (World Resources Institute, Rainforest Alliance, Proforest & Daemeter, 2018); https://data.globalforestwatch.org/documents/gfw::universal-mill-list/aboutPirker, J., Mosnier, A., Kraxner, F., Havlík, P. & Obersteiner, M. What are the limits to oil palm expansion? Glob. Environ. Change 40, 73–81 (2016).
Google Scholar 
Fischer, G. et al. Global Agro-Ecological Zones 4 (GAEZ v4) – Model Documentation (FAO, 2021); https://doi.org/10.4060/cb4744enGlobal Agro-Ecological Zoning Version 4 (GAEZ v4) (FAO & IIASA, 2021); http://www.fao.org/gaez/Tao, H. H. et al. Long-term crop residue application maintains oil palm yield and temporal stability of production. Agron. Sustain. Dev. https://doi.org/10.1007/s13593-017-0439-5 (2017).Wei, L., John Martin, J. J., Zhang, H., Zhang, R. & Cao, H. Problems and prospects of improving abiotic stress tolerance and pathogen resistance of oil palm. Plants 10, 2622 (2021).CAS 
PubMed 
PubMed Central 

Google Scholar 
Corley, R. H. & Tinker, P. B. The Oil Palm (Wiley-Blackwell, 2016).Barona, E., Ramankutty, N., Hyman, G. & Coomes, O. T. The role of pasture and soybean in deforestation of the Brazilian Amazon. Environ. Res. Lett. 5, 024002 (2010).
Google Scholar 
ten Kate, A., Kuepper, B. & Piotrowski, M. NDPE Policies Cover 83% of Palm Oil Refineries; Implementation at 78% (Chain Reaction Research, 2020); https://chainreactionresearch.com/wp-content/uploads/2020/04/NDPE-Policies-Cover-83-of-Palm-Oil-Refining-Market.pdfThe Trase Yearbook: The State Of Forest Risk Supply Chains (Trase, 2020); https://insights.trase.earth/yearbook/summaryAustin, K. G. et al. Shifting patterns of oil palm driven deforestation in Indonesia and implications for zero-deforestation commitments. Land Use Policy 69, 41–48 (2017).
Google Scholar 
Furumo, P. R., Rueda, X., Rodríguez, J. S. & Parés Ramos, I. K. Field evidence for positive certification outcomes on oil palm smallholder management practices in Colombia. J. Clean. Prod. 245, 118891 (2020).
Google Scholar 
Carlson, K. M. et al. Effect of oil palm sustainability certification on deforestation and fire in Indonesia. Proc. Natl Acad. Sci. USA 115, 121–126 (2018).CAS 
PubMed 

Google Scholar 
Heilmayr, R., Carlson, K. M. & Benedict, J. J. Deforestation spillovers from oil palm sustainability certification. Environ. Res. Lett. 15, 075002 (2020).CAS 

Google Scholar 
Impact (RSPO, 2022); https://www.rspo.org/impactBastos Lima, M. G., Persson, U. M. & Meyfroidt, P. Leakage and boosting effects in environmental governance: a framework for analysis. Environ. Res. Lett. 14, 105006 (2019).
Google Scholar 
Corley, R. H. V. How much palm oil do we need? Environ. Sci. Policy https://doi.org/10.1016/j.envsci.2008.10.011 (2009).FAOSTAT: Food and Agriculture Data (FAO, 2020); https://www.fao.org/faostat/en/Olson, D. M. et al. Terrestrial ecoregions of the world: a new map of life on Earth: a new global map of terrestrial ecoregions provides an innovative tool for conserving biodiversity. BioScience 51, 933–938 (2001).
Google Scholar 
Murphy, B. P., Andersen, A. N. & Parr, C. L. The underestimated biodiversity of tropical grassy biomes. Phil. Trans. R. Soc. B 371, 20150319 (2016).PubMed 
PubMed Central 

Google Scholar 
Smith, J. R., Hendershot, J. N., Nova, N. & Daily, G. C. The biogeography of ecoregions: descriptive power across regions and taxa. J. Biogeogr. https://doi.org/10.1111/jbi.13871 (2020).Klink, C. A. & Machado, R. B. Conservation of the Brazilian Cerrado. Conserv. Biol. 19, 707–713 (2005).
Google Scholar 
Strassburg, B. B. N. et al. Moment of truth for the Cerrado hotspot. Nat. Ecol. Evol. https://doi.org/10.1038/s41559-017-0099 (2017).le Polain de Waroux, Y. et al. The restructuring of South American soy and beef production and trade under changing environmental regulations. World Dev. 121, 188–202 (2019).
Google Scholar 
Nepstad, L. S. et al. Pathways for recent Cerrado soybean expansion: extending the soy moratorium and implementing integrated crop livestock systems with soybeans. Environ. Res. Lett. 14, 044029 (2019).
Google Scholar 
Searchinger, T. D. et al. High carbon and biodiversity costs from converting Africa’s wet savannahs to cropland. Nat. Clim. Change https://doi.org/10.1038/nclimate2584 (2015).Cardoso Da Silva, J. M. & Bates, J. M. Biogeographic patterns and conservation in the South American Cerrado: a tropical savanna hotspot. BioScience 52, 225–233 (2002).
Google Scholar 
Poggio, L. et al. SoilGrids 2.0: producing soil information for the globe with quantified spatial uncertainty. SOIL 7, 217–240 (2021).Hill, T. C., Williams, M., Bloom, A. A., Mitchard, E. T. A. & Ryan, C. M. Are inventory based and remotely sensed above-ground biomass estimates consistent? PLoS ONE https://doi.org/10.1371/journal.pone.0074170 (2013).Ryan, C. M. et al. Ecosystem services from southern African woodlands and their future under global change. Phil. Trans. R. Soc. B https://doi.org/10.1098/rstb.2015.0312 (2016).Grace, J., Jose, J. S., Meir, P., Miranda, H. S. & Montes, R. A. Productivity and carbon fluxes of tropical savannas. J. Biogeogr. 33, 387–400 (2006).
Google Scholar 
Scharlemann, J. P., Tanner, E. V., Hiederer, R. & Kapos, V. Global soil carbon: understanding and managing the largest terrestrial carbon pool. Carbon Manag. 5, 81–91 (2014).CAS 

Google Scholar 
Quezada, J. C., Etter, A., Ghazoul, J., Buttler, A. & Guillaume, T. Carbon neutral expansion of oil palm plantations in the Neotropics. Sci. Adv. 5, eaaw4418 (2019).CAS 
PubMed 
PubMed Central 

Google Scholar 
Aleman, J. C., Blarquez, O. & Staver, C. A. Land-use change outweighs projected effects of changing rainfall on tree cover in sub-Saharan Africa. Glob. Change Biol. https://doi.org/10.1111/gcb.13299 (2016).Espírito-Santo, M. M. et al. Understanding patterns of land-cover change in the Brazilian Cerrado from 2000 to 2015. Phil. Trans. R. Soc. B https://doi.org/10.1098/rstb.2015.0435 (2016).Overbeck, G. E. et al. Conservation in Brazil needs to include non-forest ecosystems. Divers. Distrib. https://doi.org/10.1111/ddi.12380 (2015).Hoekstra, J. M., Boucher, T. M., Ricketts, T. H. & Roberts, C. Confronting a biome crisis: global disparities of habitat loss and protection. Ecol. Lett. https://doi.org/10.1111/j.1461-0248.2004.00686.x (2005).RTRS Standard for Responsible Soy Production Version 3.1 (RTRS, 2017); https://responsiblesoy.org/wp-content/uploads/2019/08/RTRS%20Standard%20Responsible%20Soy%20production%20V3.1%20ING-LOW.pdfBatlle-Bayer, L., Batjes, N. H. & Bindraban, P. S. Changes in organic carbon stocks upon land use conversion in the Brazilian Cerrado: a review. Agric. Ecosyst. Environ. 137, 47–58 (2010).CAS 

Google Scholar 
Rockström, J., Falkenmark, M., Lannerstad, M. & Karlberg, L. The planetary water drama: dual task of feeding humanity and curbing climate change. Geophys. Res. Lett. 39, LXXXXX (2012).
Google Scholar 
Ocampo-Peñuela, N., Garcia-Ulloa, J., Ghazoul, J. & Etter, A. Quantifying impacts of oil palm expansion on Colombia’s threatened biodiversity. Biol. Conserv. https://doi.org/10.1016/j.biocon.2018.05.024 (2018).Gilroy, J. J. et al. Minimizing the biodiversity impact of Neotropical oil palm development. Glob. Change Biol. 21, 1531–1540 (2015).
Google Scholar 
Bonn Challenge 2020 Report (IUCN, 2020); https://www.bonnchallenge.org/resources/bonn-challenge-2020-reportGilroy, J. J. et al. Cheap carbon and biodiversity co-benefits from forest regeneration in a hotspot of endemism. Nat. Clim. Change 4, 503–507 (2014).
Google Scholar 
Evans, M. C. et al. Carbon farming via assisted natural regeneration as a cost-effective mechanism for restoring biodiversity in agricultural landscapes. Environ. Sci. Policy 50, 114–129 (2015).CAS 

Google Scholar 
Hunter, M. C., Smith, R. G., Schipanski, M. E., Atwood, L. W. & Mortensen, D. A. Agriculture in 2050: recalibrating targets for sustainable intensification. BioScience https://doi.org/10.1093/biosci/bix010 (2017).Beyer, R. & Rademacher, T. Species richness and carbon footprints of vegetable oils: can high yields outweigh palm oil’s environmental impact? Sustainability 13, 1813 (2021).
Google Scholar 
Lee, J. S. H., Ghazoul, J., Obidzinski, K. & Koh, L. P. Oil palm smallholder yields and incomes constrained by harvesting practices and type of smallholder management in Indonesia. Agron. Sustain. Dev. 34, 501–513 (2014).
Google Scholar 
Murphy, D. J. The future of oil palm as a major global crop: opportunities and challenges. J. Oil Palm Res. 26, 1–24 (2014).
Google Scholar 
Giam, X., Koh, L. P. & Wilcove, D. S. Tropical crops: cautious optimism. Science https://doi.org/10.1126/science.346.6212.928-a (2014).Villoria, N. B., Golub, A., Byerlee, D. & Stevenson, J. Will yield improvements on the forest frontier reduce greenhouse gas emissions? A global analysis of oil palm. Am. J. Agric. Econ. 95, 1301–1308 (2013).
Google Scholar 
Koh, L. P. & Lee, T. M. Sensible consumerism for environmental sustainability. Biol. Conserv. https://doi.org/10.1016/j.biocon.2011.10.029 (2012).Fick, S. E. & Hijmans, R. J. WorldClim 2: new 1-km spatial resolution climate surfaces for global land areas. Int. J. Climatol. 37, 4302–4315 (2017).
Google Scholar 
Harris, N., Goldman, E. & Gibbes, S. Spatial Database of Planted Trees (SDPT) Version 1.0 (World Resources Institute, 2019); https://data.globalforestwatch.org/datasets/tree-plantationsSutanudjaja, E. H. et al. PCR-GLOBWB 2: a 5 arcmin global hydrological and water resources model. Geosci. Model Dev. 11, 2429–2453 (2018).
Google Scholar 
Global Land Cover (Copernicus, 2019); https://lcviewer.vito.be/Tsendbazar, N.-E. et al. Copernicus Global Land Operations ‘Vegetation and Energy’ ‘CGLOPS−1’ Validation Report. Moderate Dynamic Land Cover 100m Version 2 (WUR, 2019); https://land.copernicus.eu/global/sites/cgls.vito.be/files/products/CGLOPS1_VR_LC100m-V2.0_I1.00.pdfSantoro, M. et al. GlobBiomass – global datasets of forest biomass. PANGAEA https://doi.org/10.1594/PANGAEA.894711 (2018).Santoro, M. et al. A detailed portrait of the forest aboveground biomass pool for the year 2010 obtained from multiple remote sensing observations. Geophys. Res. Abstr. https://doi.org/10.1002/joc.5086 (2018).Hansen, M. C. et al. High-resolution global maps of 21st-century forest cover change. Science https://doi.org/10.1126/science.1244693 (2013).Gumbricht, T. et al. Tropical and Subtropical Wetlands Distribution Version 7 (CIFOR, 2017); https://doi.org/10.17528/CIFOR/DATA.00058The IUCN Red List of Threatened Species Version 2018−1 (IUCN, 2018); https://www.iucnredlist.orgBird Species Distribution Maps of the World Version 6.0 (BirdLife International & Handbook of the Birds of the World, 2016); http://datazone.birdlife.org/species/requestdisR Core Team. R: A Language and Environment for Statistical Computing (R Foundation for Statistical Computing, 2018).Silalertruksa, T. et al. Environmental sustainability of oil palm cultivation in different regions of Thailand: greenhouse gases and water use impact. J. Clean. Prod. 167, 1009–1019 (2017).CAS 

Google Scholar 
Fourcade, Y., Engler, J. O., Rödder, D. & Secondi, J. Mapping species distributions with Maxent using a geographically biased sample of presence data: a performance assessment of methods for correcting sampling bias. PLoS ONE 9, e97122 (2014).PubMed 
PubMed Central 

Google Scholar 
Liu, Z. et al. Shifts in the extent and location of rice cropping areas match the climate change pattern in China during 1980–2010. Reg. Environ. Change https://doi.org/10.1007/s10113-014-0677-x (2015).Singh, K., McClean, C. J., Büker, P., Hartley, S. E. & Hill, J. K. Mapping regional risks from climate change for rainfed rice cultivation in India. Agric. Syst. 156, 76–84 (2017).PubMed 
PubMed Central 

Google Scholar 
Estes, L. D. et al. Comparing mechanistic and empirical model projections of crop suitability and productivity: implications for ecological forecasting. Glob. Ecol. Biogeogr. https://doi.org/10.1111/geb.12034 (2013).Thuiller, W., Georges, D., Engler, R. & Breiner, F. biomod2: Ensemble Platform for Species Distribution Modelling (2016).Hernandez, P. A., Graham, C. H., Master, L. L. & Albert, D. L. The effect of sample size and species characteristics on performance of different species distribution modeling methods. Ecography 29, 773–785 (2006).
Google Scholar 
Merow, C., Smith, M. J., Silander, J. A., Merow, C. & Silander, J. A. A practical guide to Maxent for modeling species’ distributions: what it does, and why inputs and settings matter. Ecography 36, 1058–1069 (2013).
Google Scholar 
Phillips, S. J., Anderson, R. P. & Schapire, R. E. Maximum entropy modeling of species geographic distributions. Ecol. Modell. https://doi.org/10.1016/j.ecolmodel.2005.03.026 (2006).VanDerWal, J., Shoo, L. P., Graham, C. & Williams, S. E. Selecting pseudo-absence data for presence-only distribution modeling: how far should you stray from what you know? Ecol. Modell. 220, 589–594 (2009).
Google Scholar 
Hirzel, A. H., Le Lay, G., Helfer, V., Randin, C. F. & Guisan, A. Evaluating the ability of habitat suitability models to predict species presences. Ecol. Modell. 199, 142–152 (2006).
Google Scholar 
Engler, R., Guisan, A. & Rechsteiner, L. An improved approach for predicting the distribution of rare and endangered species from occurrence and pseudo-absence data. J. Appl. Ecol. https://doi.org/10.1111/j.0021-8901.2004.00881.x (2004).Allouche, O., Tsoar, A. & Kadmon, R. Assessing the accuracy of species distribution models: prevalence, kappa and the true skill statistic (TSS). J. Appl. Ecol. https://doi.org/10.1111/j.1365-2664.2006.01214.x (2006).Global Spatially-Disaggregated Crop Production Statistics Data for 2010 Version 1.1. (International Food Policy Research Institute, 2019); https://doi.org/10.7910/DVN/PRFF8V/M2EMBNHofste, R. W. et al. Aqueduct 3.0: Updated Decision-Relevant Global Water Risk Indicators (World Resources Institute, 2019); https://www.wri.org/research/aqueduct-30-updated-decision-relevant-global-water-risk-indicatorsCarr, M. K. V. The water relations and irrigation requirements of oil palm (Elaeis guineensis): a review. Exp. Agric. 47, 629–652 (2011).
Google Scholar 
Yusop, Z., Hui, C. M., Garusu, G. J. & Katimon, A. Estimation of evapotranspiration in oil palm catchments by short-time period water-budget method. Malays. J. Civ. Eng. 20, 160–174 (2008).
Google Scholar 
Hargreaves, G. H. & Allen, R. G. History and evaluation of Hargreaves evapotranspiration equation. J. Irrig. Drain. Eng. 129, 53–63 (2003).
Google Scholar 
Trabucco, A. & Zomer, R. J. Global aridity index and potential evapotranspiration (ET0) climate database v2. Figshare https://doi.org/10.6084/m9.figshare.7504448.v3 (2019).Protected Planet: The World Database on Protected Areas (WDPA) (UNEP-WCMC & IUCN, 2020); www.protectedplanet.net/en/thematic-areas/wdpaDudley, N. (ed.) Guidelines for Applying Protected Area Management Categories (IUCN, 2008).Juffe-Bignoli, D. et al. World Database on Protected Areas User Manual 1.5 (UNEP-WCMC, 2017); https://www.protectedplanet.net/en/resources/wdpa-manualChave, J. J. et al. Tree allometry and improved estimation of carbon stocks and balance in tropical forests. Oecologia 145, 87–99 (2005).CAS 
PubMed 

Google Scholar 
Jetz, W., Wilcove, D. S. & Dobson, A. P. Projected impacts of climate and land-use change on the global diversity of birds. PLoS Biol. https://doi.org/10.1371/journal.pbio.0050157 (2007).Beyer, R. M. & Manica, A. Historical and projected future range sizes of the world’s mammals, birds, and amphibians. Nat. Commun. 11, 5633 (2020).CAS 
PubMed 
PubMed Central 

Google Scholar 
Beyer, R. M. & Manica, A. Global and country-level data of the biodiversity footprints of 175 crops and pasture. Data Brief 36, 106982 (2021).CAS 
PubMed 
PubMed Central 

Google Scholar 
Cobertura de la Tierra 100K Periodo 2018 (IDEAM, Instituto de Hidrología, Meteorología y Estudios Ambientales, 2021); http://www.siac.gov.co/catalogo-de-mapasSouza, C. M. et al. Reconstructing three decades of land use and land cover changes in Brazilian biomes with Landsat archive and Earth Engine. Remote Sens. 12, 2735 (2020).
Google Scholar 
MapBiomas Project – Collection 6 of the Annual Series of Land Use and Land Cover Maps of Brazil (MapBiomas, 2021); https://plataforma.brasil.mapbiomas.org/Veldman, J. W. & Putz, F. E. Grass-dominated vegetation, not species-diverse natural savanna, replaces degraded tropical forests on the southern edge of the Amazon Basin. Biol. Conserv. https://doi.org/10.1016/j.biocon.2011.01.011 (2011).Portillo-Quintero, C. A. & Sánchez-Azofeifa, G. A. Extent and conservation of tropical dry forests in the Americas. Biol. Conserv. https://doi.org/10.1016/j.biocon.2009.09.020 (2010).Veldman, J. W. et al. Toward an old-growth concept for grasslands, savannas, and woodlands. Front. Ecol. Environ. https://doi.org/10.1890/140270 (2015).Zaloumis, N. P. & Bond, W. J. Reforestation or conservation? The attributes of old growth grasslands in South Africa. Phil. Trans. R. Soc. B https://doi.org/10.1098/rstb.2015.0310 (2016).Garnett, S. T. et al. A spatial overview of the global importance of Indigenous lands for conservation. Nat. Sustain. https://doi.org/10.1038/s41893-018-0100-6 (2018).Djoudi, H., Vergles, E., Blackie, R. R., Koame, C. K. & Gautier, D. Dry forests, livelihoods and poverty alleviation: understanding current trends. Int. For. Rev. https://doi.org/10.1505/146554815815834868 (2015).Ground-Truthing to Improve Due Diligence on Human Rights in Deforestation-Risk Supply Chains (Forest Peoples Programme, 2020); https://www.forestpeoples.org/en/ground-truthing-to-improve-due-diligenceDrost, S., Rijk, G. & Piotrowski, M. Oil Palm Growers Exposed to USD 0.4-5.9B in Social Compensation Risk (Chain Reaction Research, 2019); https://chainreactionresearch.com/wp-content/uploads/2019/12/Social-compensation-risks-for-palm-growers-4.pdf More