Ecology

To build Sentinel2GlobalLULC, we followed two main steps. First, we established a spatio-temporal consensus between 15 global LULC products for 29 LULC classes. Then, we extracted the maximum number of Sentinel-2 RGB images representing each class. Each image is a tile that has 224 × 224 pixels at 10 × 10 m spatial resolution and was built as a cloud-free composite from all the Sentinel-2 images acquired between June 2015 and October 2020. Both tasks were implemented using GEE, an efficient programming, processing and visualisation platform that allowed us to have free manipulation and access to all used LULC products and Sentinel-2 imagery, simultaneously.Finding spatio-temporal agreement across 15 global LULC productsTo establish the spatio-temporal consensus between different LULC products for each one of the 29 LULC classes, we followed four steps: (1)Identification of the LULC products to be used in the consensus, (2)Standardization and harmonization of the LULC legend that was subsequently used to annotate the image tiles, (3)Spatio-temporal aggregation across LULC products, and (4)Spatial reprojection and tile selection based on optimized spatial purity thresholds.Global LULC products selectionThe adopted purity measure for spatio-temporal agreement across the 15 global LULC products we selected from GEE (Table 2) aims to find areas of high consensus to maximize the annotation quality. Spatial and temporal consensus across such rich diversity of LULC products, in terms of spatial resolution, time coverage, satellite source, LULC classes and accuracy, was used as a source of robustness for our subsequent LULC annotation. Products outside GEE were not used due to computing limitations.Table 2 Main characteristics of the 15 global Land-Use and Land-Cover (LULC) products available in Google Earth Engine (GEE) that were combined to find consensus in the global distribution of 29 main LULC classes.Full size tableStandardization and Harmonization of LULC legendsLand cover (LC) data describes the main type of natural ecosystem that occupies an area; either by vegetation types such as shrublands, grasslands and forests, or by other biophysical classes such as permanent snow, bare land and water bodies. Land use (LU) includes the way in which humans modify or exploit an area, such as urban areas or agricultural fields.To build our 29 LULC classes nomenclature, we established a standardization and harmonization approach based on expert knowledge. During this process, we took into account both the needs of different practitioners in the global and regional LULC mapping field and the thematic resolution of the global LULC legends available in GEE. Our nomenclature consists of 23 LC and 6 LU distinct classes identified through specific consensus rules across 15 LULC products (see Table 4). A six-level (L0 to L5) hierarchical structure was adopted in the creation of these 29 LULC classes (Fig. 2). To facilitate the inter-operability of our 29 legends at the finest level L5 across all LULC products and with the widely used FAO’s hierarchical Land Cover Classification System (LCCS)1, we have established an LULC classification system where the 29 classes can be mapped directly to FAO’s LCCS as explained in the table of Supplementary File 1. The LC part in our dataset contains 20 terrestrial ecosystems and 3 aquatic ecosystems. The terrestrial systems are: Barren lands, Grasslands, Permanent snow, Moss and Lichen lands, Close shrublands, Open shrublands, in addition to 12 Forests classes that differed in their tree cover, phenology, and leaf type. The aquatic classes are: Marine water bodies, Continental water bodies, and Wetlands; furthermore, wetlands were divided into 3 classes: Marshlands, Mangroves and Swamps. The LU part is composed of urban areas and 5 coarse cropland types that differed in their irrigation regime and leaf type. In Table 3, you can find the semantic definition of each one of the 29 classes in Sentinel2GlobalLULC. We provided a table in Supplementary File 2, for a more detailed definition of each LULC class.Fig. 2Tree representation of the six-level (L0 to L5) hierarchical structure of the Land-Use and Land-Cover (LULC) classes contained in the Sentinel2GlobalLULC dataset. Outter circular leafs represent the final or most detailed 29 LULC classes (C1 to C29) of level L5. The followed path to define each class is represented through inner ellipses that contain the names of intermediate classes at different levels between the division of the Earth’s surface (square) into LU and LC (level L0) and the final class circle (level L5). All LULC classes belong to three levels at least, except the 12 forest classes that belong to L5 only.Full size imageTable 3 Semantic signification of each one of the 29 Land Use and Land Cover (LULC) classes contained in the Sentinel2GlobalLULC dataset according to the six-level (L0 to L5) hierarchical structure.Full size tableCombining products across time and spaceFor each one of the 29 LULC classes, we combined in space and time the global LULC information among the 15 GEE LULC products. This way, each image was annotated with a LULC class only if all combined products agreed in its corresponding tile (i.e., 100% of agreement in space and time). For each product and LULC type, we first set one or more criteria to create a global mask at the native resolution of the product in which each pixel was classified as 1 or 0 depending on whether it met the criteria for belonging to that LULC type or not, respectively (see first stage in Table 4). For certain LULC classes, some products did not provide any relevant information, so they were not used. For example (Table 4), in Grasslands (C3), Open Shrublands (C4) and Close Shrublands (C5), we combined 14 products, while in UrbanBlUpArea (C29) and Permanent Snow (C23) we only combined 10 and 7 products, respectively.Table 4 First stage of the rule set criteria used to find consensus across the 15 Land-Use and Land-Cover (LULC) products available in Google Earth Engine (GEE) for each of the 29 LULC classes contained in the Sentinel2GlobalLULC dataset.Full size tableThen (see second stage in Table 5), for each LULC type, we calculated the average of all the masks obtained from each product to create a final global probability map from all products with values ranging between 0 and 1. Value 1 meant that all products agreed to assign that pixel to a particular LULC class, while 0 meant that none of the products assigned it to that particular class (Fig. 3). These 0-to-1 values are interpreted as the spatio-temporal purity level of each pixel to belong to a particular LULC class and are provided as metadata with each image.Table 5 Second stage of the rule set criteria used to find consensus across the 15 Land-Use and Land-Cover (LULC) products available in Google Earth Engine (GEE) for each of the 29 LULC classes contained in the Sentinel2GlobalLULC dataset.Full size tableFig. 3Example of the process of building the final global probability map for one of the 29 Land-Use and Land-Cover (LULC) classes (e.g. C1: “Barren”) by means of spatio-temporal agreement of the 15 LULC products available in Google Earth Engine (GEE). The final map is normalized to values between 0 (white, i.e., areas with no presence of C1 in any product) and 1 (black spots, i.e., areas containing or compatible with the presence of C1 in all 15 products), whereas the shades of grey corresponds to the values in between (i.e., areas that did not contain or were not compatible with the presence of C1 in some of the products). This process is divided into two stages: the first stage (the blue part, see details in Table 4) and the second stage (the yellow part, see details in Table 5). LULC products available for several years are represented with superposed rectangles, while single year products are represented with single rectangles. GMP: global probability map, NA: Not Available.Full size imageAs an example of the first stage (see details in Table 4), to specify if a given pixel belongs to Dense Evergreen Needleleaf Forest, we evaluated its tree cover level using “ ≤ “ and “ ≥ “, while for bands containing the leaf type information, we used the equal operator “ = “. For the spatio-temporal combination of multiple criteria we have used the following operators: “AND”,“OR” and “ADD”. For example, we combined the tree cover percentage criteria with the leaf type criteria using “AND” to select forest pixels that met both conditions. To combine many years instances of the same product, we used “ADD”, except for product P13, where we used “AND” to identify permanent water areas only. Whenever we used the “ADD” operator, we normalized pixel values afterwards to bring it back to a probability interval between 0 and 1 using the division by the total number of combined years or criteria.In the second stage (see details in Table 5), we combined for each LULC class the 15 global probability maps previously derived from each product to create a final global probability map (Fig. 3). This combination was carried out using various operators such as “ADD”, “MULTIPLY” and “OR”, depending on the LULC type. When “ADD” was used, the final pixel values were normalized by dividing the final addition value of each pixel by the total number of added products. The “MULTIPLY” operator was mostly used at the end, to remove urban areas from non-urban LULC classes, or to remove water from non-water LULCs. The multiplication operator was also adopted to make sure that a certain criteria was respected in the final probability map. For instance, for the swamp class, we multiplied all pixels in the final stage by a water mask where saline water areas have a value of 0 in order to eliminate mangrove from swamp pixels and vice versa. Finally, we used “OR” operator between different water related products to take advantage of the fact that they complement each other in terms of spatial-temporal coverage and accuracy.In GEE, when two products are aggregated using “ADD”, “MULTIPLY” or any other operator, the output is aggregated at the spatial resolution of the product at the left of the operator. Hence, to maintain the finest spatial resolution in the final probability map, we multiplied everything by product P15 and placed it at the left of the final “MULTIPLY” operation (See Table 5). Hence, all the 29 final probability maps were generated at the P15 spatial resolution of 30 m/pixel (except the urban class C29 which maintained the 30 m/pixel resolution of product P14).Re-projection and Selection of purity thresholdSince our objective was finding pure Sentinel-2 image tiles of 224 × 224 10-m pixels representing each LULC class, we reprojected the 30 m/pixel probability maps to 2240 m/pixel using the spatial mean reducer in GEE. That is, each pixel value at 2240 m resolution was computed using the mean over all the 30m-pixel values contained within it. Hence, the resulting pixel values at 2240 m resolution represent the purity level that each Sentinel-2 image tile of 224 × 224 10-m pixels has. We illustrated the reprojection and selection processes in Fig. 4.Fig. 4Example of the workflow to obtain a Sentinel-2 image tile of 2240 × 2240 m for one of the 29 Land-Use and Land-Cover (LULC) classes (e.g. C1: “Barren”). The process starts with the reprojected final global probability map obtained from stage two (Table 5) and ends with its exportation to the repository of a Sentinel-2 image tile of 224 × 224 pixels. The white rectangle is the only one having a probability value of 1 (Recall that the purity threshold used for Barren was 1, i.e., 100%). The black pixels has a null probability value, while the probability values between 0 and 1 are represented in gray scale levels.Full size imageFor each one of the reprojected maps, we defined a pixel value threshold to decide whether a given 2240 × 2240 m tile was representative of each LULC class or not. Since training DL image classification models needs a large number of high quality (both in terms of image quality and annotation quality) image tiles to reach a good accuracy, when the spatial purity of 100% (full agreement across products in all the pixels of the 224 × 224 tile) resulted in a small number of agreement tiles for a particular class, the purity threshold was decreased for that class until the number of tiles was larger than 1000 or further decreased in less abundant classes to a minimum of 75% of purity. The found purity value is always provided as metadata for each image in the dataset, so the user can always restrict its analysis to those image tiles and classes at any desired purity level. Decreasing the purity threshold down to 75% for the less abundant classes (e.g swamp, mangrove, etc.) was a trade-off between maintaining a good data annotation quality and providing a sufficient number of tiles in each class. In Table 6, we present the number of agreement tiles found at different purity thresholds ranging from 75% to 100% for each LULC class. This spatial purity was not further decreased since machine learning image classification models are known to be robust when the target class is spatially dominant in each training image (it occupies more than 60% of the pixels in the scene)42. On the other hand, when the number of pure tiles for a LULC class was too large to be downloaded (i.e., greater than 14000), we applied a selection algorithm as described in the Supplementary File 3, to download a maximum of 14000 spatially representative images. For this, the world was divided into a one-degree squared cell grid. If a cell contained less than 50 image tiles, we selected them all. If it contained more than 50, we applied that automatic maximum geographic distance algorithm that selected images as far from each other as possible in a number proportional to the number of existing images in that cell. The map in Fig. 6 shows the global distribution of the selected 194877 image tiles contained in Sentinel2GlobalLULC and distributed in 29 LULC classes.Table 6 Summary of the varying number of found and eventually selected Sentinel-2 image tiles of 224 × 224 pixels depending on the different consensus level reached across the 15 Land-Use and Land-Cover (LULC) products available in Google Earth Engine (GEE) for each of the 29 LULC classes contained in the Sentinel2GlobalLULC dataset.Full size tableData extractionSentinel2GlobalLULC provides the user with two types of data: Sentinel-2 RGB images (jpeg and geotif versions) and CSV files with associated metadata. In the following subsections, we describe the process for associating metadata, including the Global Human Modification (GHM) index.Global human modification index extractionAs an additional metadata related to the level of human influence in each image, we calculated for each tile in GEE, the spatial mean of the global human modification index for terrestrial lands43, where 0 means no human modification and 1 means complete transformation. Since the original GHM product was mapped at 1 × 1 km resolution, we reprojected it to 2240 × 2240 m using the same reprojection procedure explained in (Re-projection and Selection of purity threshold).CSV files generationOnce the tiles were selected, for each LULC class we listed the image tiles in descendent order of purity. Metadata included: geographical coordinates of each tile centroid, tile purity value, name and ID of the LULC class, and average GHM index for that tile. Then, we used the geographical coordinates of each tile to identify its exact administrative address geolocation. To implement this reverse geo-referencing operation, we used a free request-unlimited python module called reverse_geocoder. This way, we assigned a country code, two levels of administrative departments, and the locality to each tile.For LULC classes that had more than 14000 pure tiles, we have released the coordinates before and after the distance-based selection in case the user wants to download more tiles or use our consensus coordinates for other purposes.Sentinel-2 RGB images exportationAfter extracting all these pieces of information and grouping them into CSV files, we went back to the geographic center coordinates of each tile and used them to extract the corresponding 224 × 224 Sentinel-2 RGB tiles using GEE. Each exported image was identical to the 2240 × 2240 m area covered by its Sentinel-2 tile.We chose “Sentinel-2 MSI (Multi-Spectral Instrument) product” since it is free and publicly available in GEE at the fine resolution of 10 × 10 m. We chose “Level-1C” (i.e., top-of-atmosphere reflectance) since it provides the longest data availability of Sentinel-2 images without any modification of the data. To build RGB images, we extracted the three bands B4, B3 and B2 that correspond to Red, Green and Blue channels, respectively. More bands available in Sentinel-2 or even in Sentinel-1 images can be incorporated in the future to our dataset. However, computational limitations (i.e., the size of the dataset would be impractical) did not allowed us to handle it as a first goal. In addition, the spatial resolution of the images would be heterogeneous across bands.To minimize the inherent noise due to atmospheric conditions (e.g. clouds, aerosols, smoke, etc.) that could affect the satellite RGB images, every image was built as a temporal aggregation of all images gathered by Sentinel-2 satellites between June 2015 and October 2020. During this aggregation, only the highest quality images in the corresponding image collection were considered, as we firstly discarded all image instances where the cloud probability exceeded 20% according to the metadata provided in their corresponding Sentinel-2 collection. Then, we calculated the 25th-percentile value between all remaining images for each reflectance band (R, G, and B), and built the final image with the obtained 25-percentile values in each pixel for its RGB bands. The 25th-percentile choice was adopted giving its suitability in atmospheric noise reduction44,45,46,47,48.Usually, Sentinel-2 MSI product includes true colour images in JPEG2000 format, except for the “Level-1C” collection used here. The three original bands (B4, B3, and B2) required a saturation mapping of their reflectance values into 0–255 RGB digital values. Thus, we mapped the saturation reflectance of 3558 into 255 to obtain true RGB channels with digital values between 0 and 255. The choice of these mapping numbers was taken from the Sentinel-2 true colour image recommendations section of Sentinel user guidelines. Finally, after exporting the selected tiles for each LULC class as “.tif” images, we converted them into “.jpeg” format using a lossless conversion algorithm.Technical implementationTo implement all our methodology steps, we first created a javascript in GEE for each LULC class. Each script is a multi-task javascript where we implemented a switch command to control which task we want to execute (between the spatio-temporal aggregation task, the spatial reprojection and tiles selection task, or the data exportation task). In each one of these scripts, we selected from GEE LULC datasets repository the 15 LULC products used to build the consensus of that LULC class. Each script was responsible of elaborating the spatio-temporal combination of the selected products and generating the final consensus map for that LULC class as described in the subsection “Combining products across time and space”. Then, it exports the final global probability map as an asset into GEE server storage to make its reprojection faster. In the same script, once the consensus map exportation was done, we imported it from the GEE assets storage and reprojected it to 2240 × 2240 m resolution; then, we exported the new reprojected map into GEE assets storage again to make its analysis and processing faster. Afterwards, we imported the reprojected map into the same script and applied different processing tasks. During this processing phase, many purity threshold values were evaluated. Then, we elaborated in this same script the pure tiles identification and their center coordinates exportation into a CSV file. A distinct GEE script was developed to import, reproject and export the global GHM map. The resulted GHM map was saved as an asset too, then imported and used in each one of the 29 LULC multi-task scripts.A python script was developed separately to read the exported CSV files for each LULC class and apply the reverse geo-referencing on their pure tiles coordinates then add the found geolocalization data (country code, locality…etc) to the original CSV files as new columns. Then, another python script was implemented to read the new resulted CSV files with all their added columns (reverse geo-referencing data, GHM data) and use the center coordinates of each pure tile in that class to export first its corresponding Sentinel-2 satellite geotiff image within GEE through the python API. Finally, after downloading all the selected geotiff images from our Google drive, we created another python script to convert these geotiff images into JPEG format. More

Ahmed MZ, Li SJ, Xue X, Yin XJ, Ren SX, Jiggins FM et al. (2015) The Intracellular bacterium Wolbachia uses parasitoid wasps as phoretic vectors for efficient horizontal transmission. PLoS Pathog 11:1–19
Google Scholar 
Arai H, Hirano T, Akizuki N, Abe A, Nakai M, Kunimi Y et al. (2019) Multiple infection and reproductive manipulations of Wolbachia in Homona magnanima (Lepidoptera: Tortricidae). Microb Ecol 77:257–266PubMed 

Google Scholar 
Arai H, Lin SR, Nakai M, Kunimi Y, Inoue MN (2020) Closely related male-killing and nonmale-killing Wolbachia strains in the oriental tea tortrix Homona magnanima. Microb Ecol 79:1011–1020CAS 
PubMed 

Google Scholar 
Bailey NW, Zuk M (2008) Changes in immune effort of male field crickets infested with mobile parasitoid larvae. J Insect Physiol 54:96–104CAS 
PubMed 

Google Scholar 
Ballad JWO, Hatzidakis J, Karr TL, Kreitman M (1996) Reduced variation in Drosophila simulans mitochondrial DNA. Genetics 144:1519–1528
Google Scholar 
Birch LC (1948) The intrinsic rate of natural increase of an insect population. J Anim Ecol 17:15–26
Google Scholar 
Capobianco IIIF, Nandkumar S, Parker JD (2018) Wolbachia affects survival to different oxidative stressors dependent upon the genetic background in Drosophila melanogaster. Physiol Entomol 43:239–244
Google Scholar 
Danthanarayana W (1975) Factors determining variation in fecundity of the light brown apple moth, Epiphyas postvittana (Walker) (Tortricidae). Aust J Zool 23:309–319
Google Scholar 
Dean MD (2006) A Wolbachia-associated fitness benefit depends on genetic background in Drosophila simulans. Proc R Soc B 273:1415–1420CAS 
PubMed 
PubMed Central 

Google Scholar 
Deseo KV (1971) Study of factors influencing the fecundity and fertility of codling moth (Laspeyresia pomonella L., Lepidoptera, Tortricidae). Acta Phytopathol Hun 6:243–252
Google Scholar 
Dobson SL, Rattanadechakul W, Marsland EJ (2004) Fitness advantage and cytoplasmic incompatibility in Wolbachia single-and superinfected Aedes albopictus. Heredity 93:135–142CAS 
PubMed 

Google Scholar 
Duron O, Bouchon D, Boutin S, Bellamy L, Zhou L, Engelstädter J, Hurst GD (2008) The diversity of reproductive parasites among arthropods: Wolbachia do not walk alone. BMC Biol 6:1–12
Google Scholar 
Engelstädter J, Telschow A, Hammerstein P (2004) Infection dynamics of different Wolbachia-types within one host population. J Theor Biol 231:345–55PubMed 

Google Scholar 
Fleury F, Vavre F, Ris N, Fouillet P, Boulétreau M (2000) Physiological cost induced by the maternally-transmitted endosymbiont Wolbachia in the Drosophila parasitoid Leptopilina heterotoma. Parasitology 121:493–500PubMed 

Google Scholar 
Frank SA (1998) Dynamics of cytoplasmic incompatibility with multiple Wolbachia infections. J Theor Biol 192:213–18CAS 
PubMed 

Google Scholar 
Frank SA, Hurst LD (1996) Mitochondria and male disease. Nature 383:224–224CAS 
PubMed 

Google Scholar 
Fry AJ, Palmer MR, Rand DM (2004) Variable fitness effects of Wolbachia infection in Drosophila melanogaster. Heredity 93:379–389CAS 
PubMed 

Google Scholar 
Gómez-Valero L, Soriano-Navarro M, Pérez-Brocal V, Heddi A, Moya A, García-Verdugo JM, Latorre A (2004) Coexistence of Wolbachia with Buchnera Aphidicola and a secondary symbiont in the aphid Cinara cedri. J Bacteriol 186:6626–33PubMed 
PubMed Central 

Google Scholar 
Hoffmann AA, Turelli M, Harshman LG (1990) Factors affecting the distribution of cytoplasmic incompatibility in Drosophila simulans. Genetics 126:933–948CAS 
PubMed 
PubMed Central 

Google Scholar 
Hornett EA, Charlat S, Duplouy AMR, Davies N, Roderick GK, Wedell N et al. (2006) Evolution of male-killer suppression in a natural population. PLoS Biol 4:1643–1648CAS 

Google Scholar 
Hough JA, Pimentel D (1978) Influence of host foliage on development, survival, and fecundity of the gypsy moth. Environ Entomol 7:97–102
Google Scholar 
Ikeda T, Ishikawa H, Sasaki T (2003) Infection density of Wolbachia and level of cytoplasmic incompatibility in the Mediterranean flour moth, Ephestia kuehniella. J Invertebr Pathol 84:1–5PubMed 

Google Scholar 
Ishii T, Nakai M, Okuno S, Takatsuka J, Kunimi Y (2003) Characterization of Adoxophyes honmai single-nucleocapsid nucleopolyhedrovirus: morphology, structure, and effects on larvae. J Invertebr Pathol 83:206–214CAS 
PubMed 

Google Scholar 
Kondo N, Shimada M, Fukatsu T (2005) Infection density of Wolbachia endosymbiont affected by coinfection and host genotype. Biol Lett 1:488–491PubMed 
PubMed Central 

Google Scholar 
Lu P, Bian G, Pan X, Xi Z (2012) Wolbachia induces density-dependent inhibition to dengue virus in mosquito cells. PLoS Negl Trop D 6:1–8CAS 

Google Scholar 
Maia AHN, Luiz AJB, Campanhola C (2000) Statistical inference on associated fertility life table parameters using jackknife technique: computational aspects. J Econ Entomol 93:511–518
Google Scholar 
Mazzetto F, Gonella E, Alma A (2015) Wolbachia infection affects female fecundity in Drosophila suzukii. Bull Insectol 68:153–157
Google Scholar 
Meyer JS, Ingersoll CG, McDonald LL, Boyce MS (1986) Estimating uncertainty in population growth rates: jackknife vs. bootstrap techniques. Ecology 67:1156–1166
Google Scholar 
Moreira LA, Iturbe-Ormaetxe I, Jeffery JA, Lu G, Pyke AT, Hedges LM et al. (2009) A Wolbachia symbiont in Aedes aegypti limits infection with dengue, Chikungunya, and Plasmodium. Cell 139:1268–1278PubMed 

Google Scholar 
Mouton L, Henri H, Bouletreau M, Vavre F (2006) Effect of temperature on Wolbachia density and impact on cytoplasmic incompatibility. Parasitology 132:49–56CAS 
PubMed 

Google Scholar 
Narita S, Nomura M, Kageyama D (2007) Naturally occurring single and double infection with Wolbachia strains in the butterfly Eurema hecabe: transmission efficiencies and population density dynamics of each Wolbachia strain. FEMS Microb Ecol 61:235–245CAS 

Google Scholar 
Pigeault R, Braquart-Varnier C, Marcadé I, Mappa G, Mottin E, Sicard M (2014) Modulation of host immunity and reproduction by horizontally acquired Wolbachia. J Insect Physiol 70:125–133CAS 
PubMed 

Google Scholar 
Rancès E, Ye YH, Woolfit M, McGraw EA, O´Neill SL (2012) The relative importance of innate immune priming in Wolbachia-mediated dengue interference. PLoS Pathog 8:e1002548. https://doi.org/10.1371/journal.ppat.1002548Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
R Core Team (2021) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/Stevanovic AL, Arnold PA, Johnson KN (2015) Wolbachia -mediated antiviral protection in Drosophila larvae and adults following oral infection. Appl Environ Micro 81:8215–8223CAS 

Google Scholar 
Takamatsu T, Arai H, Abe N, Nakai M, Kunimi Y, Inoue MN (2021) Coexistence of two male-killers and their impact on the development of oriental tea tortrix Homona magnanima. Microb Ecol 81:193–202CAS 
PubMed 

Google Scholar 
Takehana A, Katsuyama T, Yano T, Oshima Y, Takada H, Aigaki T et al. (2002) Overexpression of a pattern-recognition receptor, peptidoglycan-recognition protein-LE, activates imd/relish-mediated antibacterial defense and the prophenoloxidase cascade in Drosophila larvae. Proc Natl Acad Sci USA 99:13705–13710CAS 
PubMed 
PubMed Central 

Google Scholar 
Takatsuka J, Okuno S, Ishii T, Nakai M, Kunimi Y (2010) Fitness-related traits of entomopoxviruses isolated from Adoxophyes honmai (Lepidoptera: Tortricidae) at three localities in Japan. J Invertebr Pathol 105:121–131PubMed 

Google Scholar 
Teixeira L, Ferreira Á, Ashburner M (2008) The Bacterial symbiont Wolbachia induces resistance to RNA viral infections in Drosophila melanogaster. PLoS Biol 6:e1000002. https://doi.org/10.1371/journal.pbio.1000002Article 
CAS 
PubMed Central 

Google Scholar 
Thomas P, Kenny N, Eyles D, Moreira LA, O´Neill SL, Asgari S (2011) Infection with the wMel and wMelPop strains of Wolbachia leads to higher levels of melanization in the hemolymph of Drosophila melanogaster, Drosophila simulans and Aedes aegypti. Dev Comp Immunol 35:360–365CAS 
PubMed 

Google Scholar 
Tsuruta K, Wennmann JT, Kunimi Y, Inoue MN, Nakai M (2018) Morphological properties of the occlusion body of Adoxophyes orana granulovirus. J Invertebr Pathol 154:58–64CAS 
PubMed 

Google Scholar 
Turelli M, Hoffmann AA (1991) Rapid spread of an inherited incompatibility factor in California Drosophila. Nature 353:440–442CAS 
PubMed 

Google Scholar 
Vautrin E, Vavre F (2009) Interactions between vertically transmitted symbionts: cooperation or conflict? Trends Microbiol 17:95–99CAS 
PubMed 

Google Scholar 
Vavre F, Fleury F, Lepetit D, Fouillet P, Boulétreau M (1999) Phylogenetic evidence for horizontal transmission of Wolbachia in host- parasitoid associations. Mol Biol Evol 16:1711–1723CAS 
PubMed 

Google Scholar 
Vollmer J, Schiefer A, Schneider T, Jülicher K, Johnston KL, Taylor MJ et al. (2013) Requirement of lipid II biosynthesis for cell division in cell wall-less Wolbachia, endobacteria of arthropods and filarial nematodes. Int J Med Microbiol 303:140–149CAS 
PubMed 

Google Scholar 
Voronin D, Guimarães AF, Molyneux GR, Johnston KL, Ford L, Taylor MJ (2014) Wolbachia lipoproteins: abundance, localization and serology of Wolbachia peptidoglycan associated lipoprotein and the Type IV Secretion System component, VirB6 from Brugia malayi and Aedes albopictus. Parasite Vector 7:462
Google Scholar 
Watanabe M, Miura K, Hunter MS, Wajnberg E (2011) Superinfection of cytoplasmic incompatibility-inducing Wolbachia is not additive in Orius strigicollis (Hemiptera: Anthocoridae). Heredity 106:642–648CAS 
PubMed 

Google Scholar 
Weeks AR, Turelli M, Harcombe WR, Reynolds KT, Hoffmann AA (2007) From parasite to mutualist: rapid evolution of Wolbachia in natural populations of Drosophila. PLoS Biol 5:0997–1005CAS 

Google Scholar 
Werren JH, Baldo L, Clark ME (2008) Wolbachia: Master manipulators of invertebrate biology. Nat Rev Microbiol 6:741–751CAS 
PubMed 

Google Scholar 
Xue X, Li S, Ahmed MZ, Barro PJ, Ren S, Qiu B (2012) Inactivation of Wolbachia reveals its biological roles in whitefly host. PLoS One 7:e48148. https://doi.org/10.1371/journal.pone.0048148Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Zug R, Hammerstein P (2012) Still a host of hosts for Wolbachia: analysis of recent data suggests that 40% of terrestrial arthropod species are infected. PLoS One 7:e38544. https://doi.org/10.1371/journal.pone.0038544Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Zug R, Hammerstein P (2015) Wolbachia and the insect immune system: what reactive oxygen species can tell us about the mechanisms of Wolbachia-host interactions. Front Microbiol 6:1–16
Google Scholar  More

Laboratory experimentsCulturingThe marine cyanobacterium Trichodesmium erythraeum IMS101 was obtained from the National Center for Marine Algae and Microbiota (Maine, USA) and was grown in Aquil-tricho medium prepared with 0.22 µm-filtered and microwave-sterilized oligotrophic South China Sea surface water6. The medium was enriched with various concentrations of chelexed and filter-sterilized NaH2PO4 as where indicated, and filter-sterilized vitamins and trace metals buffered with 20 µM EDTA6. The cultures were unialgal, and although they were not axenic, sterile trace metal clean techniques were applied for culturing and experimental manipulations. T. erythraeum was pre-adapted to low P condition by semi-continuously culturing at 0.5 μM PO43− and at two pCO2 levels (400 and 750 µatm) for more than one year. To start the chemostat culture, three replicates per treatment were grown in 1-L Nalgene® magnetic culture vessels (Nalgene Nunc International, Rochester, NY, USA), in which the cultures were continuously mixed by bubbling with humidified and 0.22 µm-filtered CO2–air mixtures and stirring using a suspended magnetic stir bar. The reservoirs contained Aquil-tricho medium with 1.2 μM NaH2PO4, which was delivered to the culture vessels using a peristaltic pump (Masterflex® L/S®, USA) at the dilution rate of 0.2 d−1. In all experiments, cultures were grown at ;27 °C and ~80 μmol photons m−2 s−1 (14 h:10 h light–dark cycle) in an AL-41L4 algae chamber (Percival). The concentration of Chlorophyll a (Chla) was monitored daily in the middle of the photoperiod as an indicator of biomass. When the Chla concentration remained constant for more than one generation, the system was considered to have reached steady-state, and was maintained for at least another four generations prior to sampling for further analysis.Carbonate chemistry manipulationpCO2/pH of seawater media in the culture vessels and in the reservoir was controlled by continuously bubbling with humidified and 0.22 µm-filtered CO2-air mixtures generated by CO2 mixers (Ruihua Instrument & Equipment Ltd.). During the experimental period, the pHT (pH on the total scale) of media was monitored daily using a spectrophotometric method46. The dissolved inorganic carbon (DIC) of media was analyzed by acidification and subsequent quantification of released CO2 with a CO2 analyzer (LI 7000, Apollo SciTech). Calculations of alkalinity and pCO2 were made using the CO2Sys program47, based on measurements of pHT and DIC, and the carbonate chemistry of the experiments are shown in Supplementary Table 1.Chla concentration and cell density and sizeChla concentration was measured daily following Hong et al.6. Briefly, T. erythraeum was filtered onto 3 μm polycarbonate membrane filters (Millipore), followed by heating at 65 °C for 6 min in 90% (vol/vol) methanol. After extraction the filter was removed and cell debris were spun down via centrifugation (5 min at 20,000×g) before spectrophotometric analysis. Cell density and the average cell length and width were determined at regular intervals when the chemostat cultures reached steady-state using ImageJ software. Photographs of Trichodesmium were taken using a camera (Canon DS126281, Japan) connected with an inverted microscope (Olympus CKX41, Japan). Total number and length of filaments in 1 mL of culture were measured, and the cell number of ~20 filaments was counted. The average length of cells was obtained by dividing the total length of the 20 filaments by their total cell number. The cell density of the culture was then calculated by dividing the total length of filaments in 1 mL culture by the average cell length. The average cell width was determined by measuring the width of around 1000 cells in each treatment.Elemental compositionTo determine particulate organic C (POC) and N (PON), at the end of the chemostat culturing T. erythraeum cells were collected on pre-combusted 25 mm GF/F filters (Whatman) and stored at −80 °C. Prior to analysis, the filters were dried overnight at 60 °C, treated with fuming HCl for 6 h to remove all inorganic carbon, and dried overnight again at 60 °C. After being packed in tin cups, the samples were subsequently analyzed on a PerkinElmer Series II CHNS/O Analyzer 2400.Particulate organic P (POP) was measured following Solorzano et al.48. Cells were filtered on pre-combusted 25 mm GF/F filters and rinsed twice with 2 mL of 0.17 M Na2SO4. The filters were then placed in combusted glass bottles with the addition of 2 mL of 0.017 M MgSO4, and subsequently evaporated to dryness at 95 °C and baked at 450 °C for 2 h. After cooling, 5 mL of 0.2 M HCl was added to each bottle. The bottle was then tightly capped and heated at 80 °C for 30 min, after which 5 mL Milli-Q H2O was added. Dissolved phosphate from the digested POP sample was measured colorimetrically following the standard phosphomolybdenum blue method.C uptake and N2 fixation ratesRates of short-term C uptake were determined at the end of the chemostat culturing. 100 µM NaH14CO3 (PerkinElmer) was added to 50 mL of cultures in the middle of the photoperiod, which was then incubated for 20 min under the growth conditions. After incubation, the samples were collected onto 3 μm polycarbonate membrane filters (Millipore), which were then washed with 0.22 µm-filtered oligotrophic seawater and placed on the bottom of scintillation vials. The filters were acidified to remove inorganic C by adding 500 µL of 2% HCl. The radioactivity was determined using a Tri-Carb 2800TR Liquid Scintillation Analyzer (PerkinElmer). Rates of N2 fixation (nitrogenase activity) were measured in the middle of the photoperiod for 2 h by the acetylene reduction assay49, using a ratio of 4:1 to convert ethylene production to N2 fixation.Soluble reactive phosphate (SRP) analysisWhen the chemostat cultures reached a steady-state, SRP concentrations in the culture vessels were measured at regular intervals, using the classic phosphomolybdenum blue (PMB) method with an additional step to enrich PMB on an Oasis HLB cartridge50. Briefly, 100 mL of GF/F filtered medium sample was fortified with 2 mL of ascorbic acid (100 g L−1) and 2 mL of mixed reagent (MR, the mixture of 100 mL of 130 g L−1 ammonium molybdate tetrahydrate, 100 mL of 3.5 g L−1 potassium antimony tartrate, and 300 mL of 1:1 diluted H2SO4), and then mixed completely. After standing at room temperature for 5 min, the solution was loaded onto a preconditioned Oasis HLB cartridge (3 cm3/60 mg, P/N: WAT094226, Waters Corp.) via a peristaltic pump, and then 1 mL eluent solution (0.2 M NaOH) was added to elute the sample into a cuvette, to which 0.06 mL of MR and 0.03 mL of ascorbic acid solution was added to fully develop PMB. Finally, the absorbance of PMB was measured at 700 nm using a spectrophotometer.Alkaline phosphatase (AP) activityAP activities were measured in the middle of the photoperiod using p-nitrophenylphosphate (pNPP) as a substrate51. Briefly, 5 mL of culture was incubated with 250 μL of 10 mM pNPP, 675 μL of Tris-glycine buffer (50 mM, pH 8.5) and 67.5 μL of 1 mM MgCl2 for 2 h under growth conditions. The absorbance of formed p-nitrophenol (pNP) was measured at 410 nm using a spectrophotometer.PolyP analysisAt the end of the chemostat culturing, T. erythraeum cells were filtered in the middle of the photoperiod onto 3 μm polycarbonate membrane filters (Millipore), flash frozen in liquid nitrogen, and stored at −80 °C until analysis. PolyP was quantified fluorometrically following Martin and Van Mooy22 and Martin et al.23. Briefly, samples were re-suspended in 1 mL Tris buffer (pH 7.0), sonicated for 30 s, immersed in boiling water for 5 min, sonicated for another 30 s, and then digested by 10 U DNase (Takara), RNase (2.5 U RNase A + 100 U RNase T1) (Invitrogen) and 20 μl of 20 mg mL−1 proteinase K at 37 °C for 30 min. After centrifugation for 5 min at 14,000×g, the supernatant was diluted with Tris buffer according to the range of standards curve, stained with 60 μL of 100 μM 4, 6-diamidino-2-phenylindole (DAPI) per 500 μL of samples, incubated for 7 min and then vortexed. The samples were then loaded onto a black 96-well plate and the absorption of fluorescence at an excitation wavelength of 415 nm and emission wavelength of 550 nm was measured using a PerkinElmer EnSpire® Multimode Plate Reader. PolyP standard (sodium phosphate glass Type 45) was purchased from Sigma-Aldrich. This method gives a relative measure of polyP concentration23 that is expressed as femto-equivalents of the standard per cell (feq cell−1).Cellular ATP measurementCellular ATP contents were determined when the chemostat cultures reached a steady state. T. erythraeum cells were collected in the middle of the photoperiod using an ATP Assay Kit (Beyotime Biotechnology, Shanghai, China) according to the manufacturer’s instructions. Briefly, the sample was lysed and centrifuged, and the supernatant (100 μL) was mixed with ATP detection working reagent (100 μL) and loaded onto a black 96-well plate. The luminescence was measured using a PerkinElmer EnSpire® Multimode Plate Reader.Intracellular metabolites measurementsNAD(H), NADP(H), and Glu were measured at the end of the chemostat culturing, using the liquid chromatography-tandem quadrupole mass spectrometry (LC–MS/MS) method modified from Luo et al.52. Briefly, T. erythraeum cells were gently filtered at the middle of photoperiod onto 3 μm polycarbonate membrane filters (Millipore), rapidly suspended in −80 °C precooled methanol-water (60%, v/v) mixture. After being kept in −80 °C freezer for 30 min, the sample was sonicated for 30 s, centrifuged at 12,000×g and 4 °C for 5 min, and the supernatant was filtered through a 0.2 μm filter (Jinteng®, China) and stored at −80 °C for further LC–MS/MS analysis.A 2.0 × 50 mm Phenomenex® Gemini 5u C18 110 Å column (particle size 5.2 µm, Phenomenex, USA) was used for the analysis. The mobile phases consisted of two solvents: mobile phase A (10 mM tributylamine aqueous solution, pH 4.95 with 15 mM acetic acid) and mobile phase B (100% methanol), which were delivered using an Agilent 1290 UPLC binary pump (Agilent Technologies, Palo Alto, CA, USA) at a flow rate of 200 µL min−1, with a linear gradient program implemented as follows: hold isocratic at 0% B (0–2 min); linear gradient from 0% to 85% B (2–28 min); hold isocratic at 0% B (28–34 min). The effluent from the LC column was delivered to an Agilent 6490 triple-quadrupole mass spectrometer, equipped with an electrospray ionization source operating in negative-ion mode. NAD, NADH, NADP, NADPH, and Glu were monitored in the multiple reaction monitoring modes with the transition events at m/z 662.3  > 540, 664.3  > 79, 742  > 620, 744  > 79, and 147  > 84, respectively.RNA extraction, library preparation, and sequencingAt the end of the chemostat culturing, T. erythraeum was collected in the middle of the photoperiod by filtering onto 3 μm polycarbonate membrane filters (Millipore), flash frozen in liquid nitrogen and stored at −80 °C until extraction. Total RNA was extracted using TRIzol® Reagent (Invitrogen) combined with a physical cell disruption approach by glass beads according to the manufacturer’s instructions. Genomic DNA was removed thoroughly by treating it with RNAase-free DNase I (Takara, Japan). Ribosomal RNA was removed from a total amount of 3 µg RNA using Ribo-Zero rRNA Removal kit (Illumina, USA). Subsequently, cDNA libraries were generated according to the manufacturer’s protocol of NEBNext® UltraTM Directional RNA Library Prep Kit for Illumina® (NEB, USA). The quality of the library was assessed on the Agilent Bioanalyzer 2100 system (Agilent Technologies, CA, USA). Libraries were sequenced on an Illumina Hiseq 2500 platform, yielding 136-bp paired-end reads.RNA-Seq bioinformaticsClean reads were obtained from raw data by removing reads containing adapter, ploy-N and low-quality read. Qualified sequences were mapped to the Trichodesmium erythraeum IMS101 genome (https://www.ncbi.nlm.nih.gov/nuccore/NC_008312.1) by using Bowtie2-2.2.353. Differential expression analysis for high/low pCO2 with P limitation was performed using the DESeq2 R package54. The resulting p-values were adjusted using Benjamini and Hochberg’s approach for controlling the false discovery rate. Genes with an adjusted p-value  More

The numbers are heading in the wrong direction. If the world continues on its current track, it will fall well short of achieving almost all of the 17 Sustainable Development Goals (SDGs) that the United Nations set to protect the environment and end poverty and inequality by 2030.The projected grade for:Eliminating hunger: F.Ensuring healthy lives for all: F.Protecting and sustainably using ocean resources: F.The trends were there before 2020, but then problems increased with the COVID-19 pandemic, war in Ukraine and the worsening effects of climate change. The world is in “a new uncertainty complex”, says economist Pedro Conceição, lead author of the United Nations Human Development Report.One measure of this is the drastic change in the Human Development Index (HDI), which combines educational outcomes, income and life expectancy into a single composite indicator. After 2019, the index has fallen for two successive years for the first time since its creation in 1990. “I don’t think this is a one-off, or a blip. I think this could be a new reality,” Conceição says.UN secretary-general António Guterres is worried. “We need an urgent rescue effort for the SDGs,” he wrote in the foreword to the latest progress report, published in July. Over the past year, Guterres and the heads of big UN agencies, such as the Statistics Division and the UN Development Programme, have been assessing what’s gone wrong and what needs to be done. They’re converging on the idea that it’s time to stop using gross domestic product (GDP) as the world’s main measure of prosperity, and to complement it with a dashboard of indicators, possibly ones linked to the SDGs. If this happens, it would be the biggest shift in how economies are measured since nations first started using GDP in 1953, almost 70 years ago1.
Get the Sustainable Development Goals back on track
Guterres’s is the latest in a crescendo of voices calling for GDP to be dropped as the world’s primary go-to indicator, and for a dashboard of metrics instead. In 2008, then French president Nicolas Sarkozy endorsed such a call from a team of economists, including Nobel laureates Amartya Sen and Joseph Stiglitz.And in August, the White House announced a 15-year plan to develop a new summary statistic that would show how changes to natural assets — the natural wealth on which economies depend — affect GDP. The idea, according to the project’s main architect, economist Eli Fenichel at the White House Office of Science and Technology Policy, is to help society to determine whether today’s consumption is being accomplished without compromising the future opportunities that nature provides. “GDP only gives a partial and — for many common uses — an incomplete, picture of economic progress,” Fenichel says.The fact that Guterres has made this a priority, amid so many major crises, is a sign that “going beyond GDP has been picked up at the highest level”, says Stefan Schweinfest, the director of the UN Statistics Division, based in New York City.Grappling with growth GDP is a measure of economic activity that has ended up becoming the world’s main index for economic progress. By a commonly used definition, it is the numerical sum of countries’ consumer and government spending and their business investments, adding the value of exports minus imports. When governments and businesses talk, as they regularly do, about boosting ‘economic growth’, what they mean is boosting GDP.But GDP is more than a growth target. It is also the benchmark for how countries measure themselves against each other (see ‘Growth gaps’). The United States is the world’s largest economy, as measured by GDP. China, currently second, is on a path to overtake it.

Source: World Bank

GDP also matters greatly to politicians. When India leapfrogged the United Kingdom to become the world’s fifth largest economy earlier this year, it made headline news. Last month, China reportedly delayed publication of its latest (and less-than-flattering) quarterly GDP figures so they would not appear during the Communist party’s national congress, at which Xi Jinping took a third term as president.“GDP is without question the superstar of indicators,” says Rutger Hoekstra, a researcher who studies sustainability metrics at Leiden University in the Netherlands and author of Replacing GDP by 2030.The problem with using GDP as a proxy for prosperity, says Hoekstra, is that it doesn’t reflect equally important indicators that have been heading in the opposite direction. Global GDP has increased exponentially since the Industrial Revolution, but this has coincided with high levels of income and wealth inequality, according to data compiled by the economist Thomas Piketty at the World Inequality Lab in Paris2. This is not a coincidence. Back in the 1950s, when countries pivoted economies to maximizing GDP, they knew it would mean “making the labourer produce more than he is allowed to consume”, as Pakistan’s then chief economist Mahbub ul Haq graphically put it3. “It is well to recognize that economic growth is a brutal, sordid process.”What is more, to boost GDP, nations need to indulge in environmentally damaging behaviour. In his 2021 report, entitled Our Common Agenda, Guterres writes: “Absurdly, GDP rises when there is overfishing, cutting of forests or burning of fossil fuels. We are destroying nature, but we count it as an increase in wealth.”This tension is apparent when it comes to the SDGs. GDP growth is associated with several SDG targets; in fact SDG 8 is about boosting growth. But GDP growth “can also come at the expense of progress towards other goals”, such as climate and biodiversity action, says environmental economist Pushpam Kumar, who directs a UN Environment Programme (UNEP) project, called the Inclusive Wealth Report, to measure sustainability and inequality. The latest report will be published next month.The one-number problemThe present effort by Guterres and his colleagues is not the first time policymakers have tried to improve on GDP. In 1990, a group of economists led by ul Haq and Sen designed the HDI. They were motivated in part by frustration that their countries’ often impressive growth rates masked more-dismal quality-of-life data, such as life expectancy or education.More recently, environment ministers have found that GDP-boosting priorities have got in the way of their SDG efforts. Carlos Manuel Rodríguez, the former environment minister of Costa Rica, says he urged his finance and economics colleagues to take account of the impact of economic development on water, soils, forests and fish. But they were concerned about possible reductions in GDP calculations, says Rodríguez, now chief executive of the Global Environment Facility, based in Washington DC. Costa Rica didn’t want to be the first country to implement such a change only to possibly see itself slide down the growth rankings as a result.

Industrial production, such as the work at this automobile plant in Japan, goes into GDP calculations.Credit: Akio Kon/Bloomberg via Getty

China’s environmental policymakers were confronted with a similar response when, in 2006, they tried to implement a plan called Green GDP4. Local authorities were asked to measure the economic cost of pollution and environmental damage, and offset that against their economic growth targets. “They panicked and the project was shelved,” says Vic Li, a political economist at the Education University of Hong Kong, who has studied the episode. “Reducing GDP would have affected their performance reviews, which needed GDP to always increase,” he says.It’s been a similar story in Italy. In 2019, then research minister Lorenzo Fioramonti helped to establish an agency, Well-being Italy, attached to the prime minister’s office. It was intended to test economic policy decisions against sustainability targets. “It was an uphill battle because the various economic ministries did not see this as a priority,” says Fioramonti, now an economist at the University of Surrey in Guildford, UK.Revising the rulesSo, can the latest attempt to complement GDP succeed? Economists and national statisticians who help to determine GDP’s rules say it will be a struggle.Guterres and his colleagues are proposing to include 10–20 indicators alongside GDP. But that’s a tough sell because countries see a lot of value (not to mention ease of use) in relying on one number. And GDP’s great success is that countries produce their own figures, according to internationally agreed rules, which allow for cross-comparison over time. “It’s not a metric compiled by Washington DC, Beijing or London,” says Schweinfest.At the same time, GDP is not something that can just be turned on or off. In each country, tracking the data that goes into calculating GDP is an industrial-scale operation involving government data as well as surveys of households and businesses.
Are there limits to economic growth? It’s time to call time on a 50-year argument
China, Costa Rica and Italy’s experiences suggest that an environment-adjusted GDP might be accepted only if every country signs up to the concept at the same time. In theory, this could happen at the point when GDP’s rules — known as the System of National Accounts — are being reviewed, an event that takes place roughly once every 15 years.The next revision to the rules is under way and is due to be completed in 2025. The final decision will be made by the UN Statistical Commission, a group of chief statisticians from different nations, together with the European Commission, the International Monetary Fund, the World Bank and the Organisation for Economic Co-operation and Development (OECD), the network of the world’s wealthy countries.Because the UN oversees this process, Guterres has some influence over the questions that the review is asking. As part of their research, national statisticians are exploring how to measure well-being and sustainability, along with improving the way the digital economy is valued. Economists Diane Coyle and Annabel Manley, both at the University of Cambridge, say that technology and data companies, which make up seven out of the global top ten firms by stock-market capitalization, are probably undervalued in national accounts5.However, according to Peter van de Ven, a former OECD statistician who is the lead editor of the GDP revision effort, some aspects of digital-economy valuation, along with putting a value on the environment, are unlikely to make it into a revised GDP formula, and will instead be part of the report’s supplementary data tables. One of the reasons, he says, is that national statisticians have not agreed on a valuation methodology for the environment. Nor is there agreement on how to value digital services such as when people use search engines or social-media accounts that (like the environment) are not bought and sold for money.Yet other economists, including Fenichel, say that there are well-established methods that economists use to value both digital and environmental goods and services. One way involves asking people what they would be willing to pay to keep or use something that might otherwise be free, such as a forest or an Internet search engine. Another method involves asking what people would be willing to accept in exchange for losing something otherwise free. Management scientists Erik Brynjolfsson and Avinash Collis, both at the Massachusetts Institute of Technology in Cambridge, did an experiment6 in which they computed the value of social media by paying people to give up using it.The value of natureEconomist Gretchen Daily at Stanford University in California says it’s not true that valuing the environment would make economies look smaller. It all depends on what you value. Daily is among the principal investigators of a project called Gross Ecosystem Product (GEP) that has been trialled across China and is now set to be replicated in other countries. GEP adds together the value of different kinds of ecosystem goods and services, such as agricultural products, water, carbon sequestration and recreational sites. The researchers found7 that in the Chinese province of Qinghai, the region’s total GEP exceeded its GDP.Although past efforts to avoid using GDP have stalled, this time could be different. It’s likely, as van de Ven says, that national statisticians will not add nature (or indeed the value of social media and Internet search) to the GDP formula. But the pressure for change is greater than at any time in the past.GDP is like a technical standard, such as the voltage of household electricity or driving on the left, says Coyle. “So if you want to switch to the right, you need to align people on the same approach. Everyone needs to agree.” More

Dubois, E. On Pithecanthropus Erectus: a transitional form between man and the apes. J. Anthropol. Inst. G. B. Irel. 25, 240–255 (1896).
Google Scholar 
von Koenigswald, G. H. R. Neue Pithecanthropus-funde, 1936-1938 : ein beitrag zur Kenntnis der Praehominiden Wetenschappelijke Mededeelingen ; no. 28 (Landsdrukkerij, Batavia, 1940).Janssen, R. et al. Tooth enamel stable isotopes of Holocene and Pleistocene fossil fauna reveal glacial and interglacial paleoenvironments of hominins in Indonesia. Quatern. Sci. Rev. 144, 145–154 (2016).ADS 

Google Scholar 
Bettis, E. A. et al. Way out of Africa: Early Pleistocene paleoenvironments inhabited by Homo erectus in Sangiran, Java. J. Hum. Evol. 56(1), 11–24 (2009).PubMed 

Google Scholar 
Huffman, O. Geologic context and age of the Perning/Mojokerto Homo erectus, East Java. J. Hum. Evol. 40(4), 353–362 (2001).PubMed 

Google Scholar 
Sarr, A.-C. et al. Subsiding Sundaland. Geology (Boulder) 47(2), 119–122 (2019).ADS 

Google Scholar 
Salles, T. et al. Quaternary landscape dynamics boosted species dispersal across Southeast Asia. Commun. Earth Environ. 2(1), 1–12 (2021).MathSciNet 

Google Scholar 
Husson, L., Boucher, F. C., Sarr, A., Sepulchre, P. & Cahyarini, S. Y. Evidence of Sundaland’s subsidence requires revisiting its biogeography. J. Biogeogr. 47(4), 843–853 (2020).Winder, I. C. et al. Evolution and dispersal of the genus Homo: A landscape approach. J. Hum. Evol. 87, 48–65 (2015).PubMed 

Google Scholar 
Carotenuto, F. et al. Venturing out safely: The biogeography of Homo erectus dispersal out of Africa. J. Hum. Evol. 95, 1–12 (2016).PubMed 

Google Scholar 
Larick, R. et al. Early Pleistocene 40Ar/39Ar ages for Bapang Formation hominins, Central Jawa, Indonesia. Proc. Natl. Acad. Sci. PNAS 98(9), 4866–4871 (2001).ADS 
PubMed 

Google Scholar 
Swisher, C. C., Curtis, G. H., Jacob, T., Getty, A. G. & Suprijo, A. Age of the earliest known hominids in Java, Indonesia. Science 263(5150), 1118–1121 (1994).ADS 
PubMed 

Google Scholar 
Sémah, F., Saleki, H., Falguŕes, C., Féraud, G. & Djubiantono, T. Did early man reach Java during the Late Pliocene?. J. Archaeol. Sci. 27(9), 763–769 (2000).
Google Scholar 
Bettis, E. et al. Landscape development preceding Homo erectus immigration into Central Java, Indonesia: The Sangiran Formation Lower Lahar. Palaeogeogr. Palaeoclimatol. Palaeoecol. 206(1), 115–131 (2004).
Google Scholar 
Matsu’ura, S. et al. Age control of the first appearance datum for Javanese Homo erectus in the Sangiran area. Science 367(6474), 210–214 (2020).ADS 
PubMed 

Google Scholar 
Granger, D. E. & Muzikar, P. F. Dating sediment burial with in situ-produced cosmogenic nuclides: Theory, techniques, and limitations. Earth Planet. Sci. Lett. 188(1), 269–281 (2001).ADS 

Google Scholar 
Shen, G., Gao, X., Gao, B. & Granger, D. E. Age of Zhoukoudian Homo erectus determined with 26Al/10Be burial dating. Nature 458(7235), 198–200 (2009).ADS 
PubMed 

Google Scholar 
Pappu, S. et al. Early Pleistocene presence of Acheulian Hominins in South India. Science 331(6024), 1596–1599 (2011).ADS 
PubMed 

Google Scholar 
Lebatard, A.-E. et al. Dating the Homo erectus bearing travertine from Kocabaş (Denizli, Turkey) at at least 1.1 Ma. Earth Planet. Sci. Lett.390, 8–18 (2014).Lebatard, A.-E., Bourlès, D. L. & Braucher, R. Absolute dating of an Early Paleolithic site in Western Africa based on the radioactive decay of in situ-produced 10Be and 26Al. Nucl. Instrum. Methods Phys. Res. Sect. B 456, 169–179 (2019).ADS 

Google Scholar 
Braucher, R., Oslisly, R., Mesfin, I., Ntoutoume Mba, P. P. & Team, A. In situ-produced 10 Be and 26 Al indirect dating of Elarmékora Earlier Stone Age artifacts: First attempt in a savannah forest mosaic in the middle Ogooué valley, Gabon. Philos. Trans. Biol. Sci. (2021) .Grimaud-Hervé, D. et al. Position of the posterior skullcap fragment from Sendang Klampok (Sangiran Dome, Java, Indonesia) among the Javanese Homo erectus record. Quatern. Int. 416, 193–209 (2016).
Google Scholar 
Sartono, S. Observations on a new skull of Pithecanthropus erectus (Pithecanthropus VIII), from Sangiran, Central Java. Koninklijke Akademie Wetenschappen te Amsterdam 74, 185–194 (1971).
Google Scholar 
Wessel, P., Smith, W. H. F., Scharroo, R., Luis, J. & Wobbe, F. Generic mapping tools: Improved version released. EOS Trans. Am. Geophys. Union 94(45), 409–410. https://doi.org/10.1002/2013EO450001 (2013).Article 
ADS 

Google Scholar 
Antón, S., Potts, R. & Aiello, L. Evolution of Early Homo: An integrated biological perspective. Science (New York, N.Y.)345 (2014). https://doi.org/10.1126/science.1236828.Luo, L. et al. The first radiometric age by isochron 26Al/10Be burial dating for the Early Pleistocene Yuanmou hominin site, southern China. Quat. Geochronol. 55, 101022. https://doi.org/10.1016/j.quageo.2019.101022 (2019).Article 

Google Scholar 
Zaim, Y. et al. New 1.5 million-year-old Homo erectus maxilla from Sangiran (Central Java, Indonesia). J. Hum. Evol.61(4), 363–376 (2011).Rizal, Y. et al. Last appearance of Homo erectus at Ngandong, Java, 117,000–108,000 years ago. Nature 577(7790), 381–385 (2020).PubMed 

Google Scholar 
McRae, B. & Beier, P. Circuit theory predicts gene flow in plant and animal populations. Proc. Natl. Acad. Sci. USA 104, 19885–90. https://doi.org/10.1073/pnas.0706568104 (2008).Article 
ADS 

Google Scholar 
Quaglietta, L. & Porto, M. SiMRiv: An R package for mechanistic simulation of individual, spatially-explicit multistate movements in rivers, heterogeneous and homogeneous spaces incorporating landscape bias. Mov. Ecol. https://doi.org/10.1186/s40462-019-0154-8 (2019).Article 
PubMed 
PubMed Central 

Google Scholar 
Landau, V. A., Shah, V. B., Anantharaman, R. & Hall, K. R. Omniscape.jl: Software to compute omnidirectional landscape connectivity. J. Open Source Softw.6(57), 2829 (2021). https://doi.org/10.21105/joss.02829.Salles, T., Mallard, C. & Zahirovic, S. gospl: Global Scalable Paleo Landscape Evolution. J. Open Source Softw.5(56), 2804 (2020). https://doi.org/10.21105/joss.02804.Husson, L. et al. Slow geodynamics and fast morphotectonics in the far East Tethys. Geochem. Geophys. Geosyst. 23(1), n/a (2022).Valdes, P., Scotese, C. & Lunt, D. Deep ocean temperatures through time. Climate Past 17, 1483–1506. https://doi.org/10.5194/cp-17-1483-2021 (2021).Article 
ADS 

Google Scholar 
Hyodo, M. et al. High-resolution record of the Matuyama–Brunhes transition constrains the age of Javanese Homo erectus in the Sangiran dome, Indonesia. Proc. Natl. Acad. Sci. PNAS 108(49), 19563–19568 (2011).ADS 
PubMed 

Google Scholar 
Brasseur, B., Sémah, F., Sémah, A.-M. & Djubiantono, T. Pedo-sedimentary dynamics of the Sangiran dome hominid bearing layers (Early to Middle Pleistocene, central Java, Indonesia): A palaeopedological approach for reconstructing ‘Pithecanthropus’ (Javanese Homo erectus) palaeoenvironment. Quatern. Int. 376, 84–100 (2015).
Google Scholar 
Falguéres, C. et al. Geochronology of early human settlements in Java: What is at stake?. Quatern. Int. 416, 5–11 (2016).
Google Scholar 
Roach, N. et al. Pleistocene footprints show intensive use of lake margin habitats by Homo erectus groups. Sci. Rep. 121 (2016). https://doi.org/10.1038/srep26374.Simandjuntak, T. O. & Barber, A. J. Contrasting tectonic styles in the Neogene orogenic belts of Indonesia. Geol. Soc. Spec. Pub. 106(1), 185–201 (1996).
Google Scholar 
Clements, B., Hall, R., Smyth, H. R. & Cottam, M. A. Thrusting of a volcanic arc; a new structural model for Java. Pet. Geosci. 15(2), 159–174 (2009).
Google Scholar 
Joordens, J., Wesselingh, F., de Vos, J., Vonhof, H. & Kroon, D. Relevance of aquatic environments for hominins: A case study from Trinil (Java, Indonesia). J. Hum. Evol. 57(6), 656–671 (2009).PubMed 

Google Scholar 
Berghuis, H. et al. Hominin homelands of East Java: Revised stratigraphy and landscape reconstructions for Plio-Pleistocene Trinil. Quatern. Sci. Rev. 260, 106912 (2021).
Google Scholar 
Fort, J., Pujol, T. & Cavalli-Sforza, L. Palaeolithic populations and waves of advance. Camb. Archaeol. J. 14, 53–61. https://doi.org/10.1017/S0959774304000046 (2004).Article 

Google Scholar 
Hamilton, M. & Buchanan, B. Spatial gradients in Clovis-age radiocarbon dates across North America suggest rapid colonization from the north. Proc. Natl. Acad. Sci. USA 104, 15625–30. https://doi.org/10.1073/pnas.0704215104 (2007).Article 
ADS 
PubMed 
PubMed Central 

Google Scholar 
Hazelwood, L. & Steele, J. Spatial dynamics of human dispersals: Constraints on modelling and archaeological validation. J. Archaeol. Sci. 31, 669–679. https://doi.org/10.1016/j.jas.2003.11.009 (2004).Article 

Google Scholar 
Bae, C., Li, F., Liuling, C., Wang, W. & Hanlie, H. Hominin distribution and density patterns in pleistocene China: Climatic influences. Palaeogeogr. Palaeoclimatol. Palaeoecol. 512 (2018). https://doi.org/10.1016/j.palaeo.2018.03.015.Timmermann, A. et al. Climate effects on archaic human habitats and species successions. Nature 604, 1–7. https://doi.org/10.1038/s41586-022-04600-9 (2022).Article 

Google Scholar 
Bailey, G. N., Reynolds, S. C. & King, G. C. Landscapes of human evolution: Models and methods of tectonic geomorphology and the reconstruction of hominin landscapes. J. Hum. Evol. 60(3), 257–280 (2011).PubMed 

Google Scholar 
Sarr, A., Sepulchre, P. & Husson, L. Impact of the Sunda Shelf on the Climate of the Maritime Continent. J. Geophys. Res. Atmos. 124(5), 2574–2588 (2019).ADS 

Google Scholar 
Louys, J. & Roberts, P. Environmental drivers of megafauna and hominin extinction in Southeast Asia. Nature 586(7829), 402–406 (2020).ADS 
PubMed 

Google Scholar 
Raia, P. et al. Past extinctions of homo species coincided with increased vulnerability to climatic change. One Earth 3(4), 480–490 (2020).ADS 

Google Scholar 
Zhu, Z. et al. Hominin occupation of the Chinese Loess Plateau since about 2.1 million years ago. Nature 559(7715), 608–612 (2018).Gabunia, L. et al. Earliest Pleistocene hominid cranial remains from Dmanisi, Republic of Georgia: Taxonomy, geological setting, and age. Science 288, 1019–1025. https://doi.org/10.1126/science.288.5468.1019 (2000).Article 
ADS 
PubMed 

Google Scholar 
Lordkipanidze, D. et al. A complete skull from Dmanisi, Georgia, and the evolutionary biology of early homo. Science 342(6156), 326–331 (2013).ADS 
PubMed 

Google Scholar 
Baba, H. et al. Homo erectus calvarium from the pleistocene of java. Sci. (Am. Assoc. Adv. Sci.) 299 (5611), 1384–1388 (2003) .Ciochon, R. L. & Bettis, E. A. III. Asian Homo erectus converges in time. Nature 458(7235), 153–154 (2009).ADS 
PubMed 

Google Scholar 
Dennell, R. & Roebroeks, W. An Asian perspective on early human dispersal from Africa. Nature 438(7071), 1099–1104 (2005).ADS 
PubMed 

Google Scholar 
Martinon-Torres, M. et al. Dental evidence on the hominin dispersals during the Pleistocene. Proc. Natl. Acad. Sci. PNAS 104(33), 13279–13282 (2007).ADS 
PubMed 

Google Scholar 
Wood, B. Did early Homo migrate “out of’’ or “in to’’ Africa?. Proc. Natl. Acad. Sci. PNAS 108(26), 10375–10376 (2011).ADS 
PubMed 

Google Scholar 
Shen, G. et al. Isochron 26Al/10Be burial dating of Xihoudu: Evidence for the earliest human settlement in northern China. Anthropologie 124, 102790. https://doi.org/10.1016/j.anthro.2020.102790 (2020).Article 

Google Scholar 
Chmeleff, J., von Blanckenburg, F., Kossert, K. & Jakob, D. Determination of the 10Be half-life by multicollector ICP-MS and liquid scintillation counting. Nucl. Instrum. Methods Phys. Res. Sect. B Beam Interact. Mater. Atoms268(2), 192–199 (2010).Korschinek, G. et al. A new value for the half-life of 10Be by Heavy-Ion Elastic Recoil Detection and liquid scintillation counting. Nucl. Instrum. Methods Phys. Res. Sect. B Beam Interact. Mater. Atoms 268(2), 187–191 (2010) .Nishiizumi, K. Preparation of 26Al AMS standards. Nucl. Inst. and Meth. in Phys. Res. 223-224, 388–392 (2004).Norris, T. L., Gancarz, A. J., Rokop, D. J. & Thomas, K. W. Half-life of 26Al. J. Geophys. Res. Solid Earth 88(S01), B331–B333 (1983).ADS 

Google Scholar 
Braucher, R., Merchel, S., Borgomano, J. & Bourlès, D. Production of cosmogenic radionuclides at great depth: A multi element approach. Earth Planet. Sci. Lett. 309(1), 1–9 (2011).ADS 

Google Scholar 
Braucher, R. et al. Preparation of ASTER in-house 10Be/9Be standard solutions. Nucl. Instrum. Methods Phys. Res. Sect. B Beam Interact. Mater. Atoms361, 335–340 (2015) .Merchel, S. & Bremser, W. First international 26Al interlaboratory comparison—Part I. Nucl. Instrum. Methods Phys. Res. 223–224, 393–400 (2004).ADS 

Google Scholar 
Arnold, M. et al. The French accelerator mass spectrometry facility ASTER: Improved performance and developments. Nucl. Instrum. Methods Phys. Res. 268(11), 1954–1959 (2010).ADS 

Google Scholar 
Borchers, B. et al. Geological calibration of spallation production rates in the CRONUS-Earth project. Quat. Geochronol. 31, 188–198 (2016).
Google Scholar 
Stone, J. O. Air pressure and cosmogenic isotope production. J. Geophys. Res. Solid Earth 105(B10), 23753–23759 (2000).
Google Scholar 
Bintanja, R. & van de Wal, R. S. W. North American ice-sheet dynamics and the onset of 100,000-year glacial cycles. Nature 454, 869–872. https://doi.org/10.1038/nature07158 (2008).Article 
ADS 
PubMed 

Google Scholar 
Field, J. & Mirazon Lahr, M. Assessment of the Southern Dispersal: GIS-Based Analyses of Potential Routes at Oxygen Isotopic Stage 4. J. World Prehist. 19, 1–45 (2005). https://doi.org/10.1007/s10963-005-9000-6.Howey, M. Multiple pathways across past landscapes: Circuit theory as a complementary geospatial method to least cost path for modeling past movement. J. Archaeol. Sci. 38, 2523–2535. https://doi.org/10.1016/j.jas.2011.03.024 (2011).Article 

Google Scholar 
Tassi, F. et al. Early modern human dispersal from Africa: Genomic evidence for multiple waves of migration. Investig. Genet. 6, 13. https://doi.org/10.1186/s13323-015-0030-2 (2015).Article 
PubMed 
PubMed Central 

Google Scholar 
Kealy, S., Louys, J. & O’Connor, S. Least-cost pathway models indicate northern human dispersal from Sunda to Sahul. J. Hum. Evol. 125, 59–70. https://doi.org/10.1016/j.jhevol.2018.10.003 (2018).Article 
PubMed 

Google Scholar 
Dennell, R. W., Rendell, H. M. & Hailwood, E. Late pliocene artefacts from northern Pakistan. Curr. Anthropol. 29(3), 495–498 (1988).
Google Scholar 
Zhu, R. et al. Early evidence of the genus homo in east asia. J. Hum. Evol. 55(6), 1075–1085 (2008).PubMed 

Google Scholar 
Gowen, K. M. & de Smet, T. S. Testing least cost path (LCP) models for travel time and kilocalorie expenditure: Implications for landscape genomics. PLoS ONE 15(9), 1–20. https://doi.org/10.1371/journal.pone.0239387 (2020).Article 

Google Scholar 
Walt, S. et al. scikit-image: Image processing in Python. PeerJ 2, e453. https://doi.org/10.7287/peerj.preprints.336v2 (2014).Article 
PubMed 
PubMed Central 

Google Scholar 
Mueller, T. & Fagan, W. Search and navigation in dynamic environments—From individual behaviors to population distributions. Oikos 117, 654–664. https://doi.org/10.1111/j.0030-1299.2008.16291.x (2008).Article 

Google Scholar 
Bastille-Rousseau, G., Douglas-Hamilton, I., Blake, S., Northrup, J. & Wittemyer, G. Applying network theory to animal movements to identify properties of landscape space use. Ecol. Appl. 28 (2018). https://doi.org/10.1002/eap.1697.Michelot, T., Langrock, R. & Patterson, T. moveHMM: An R package for the statistical modelling of animal movement data using hidden Markov models. Methods Ecol. Evol. 7 (2016). https://doi.org/10.1111/2041-210X.12578 .Benhamou, S. How many animals really do the Lévy Walk. Ecology 88, 1962–9. https://doi.org/10.1890/06-1769.1 (2007).Article 
PubMed 

Google Scholar 
Turchin, P. Quantitative Analysis of Movement: Measuring and Modeling Population Redistribution of Plants and Animals (Sinauer Associates, Sunderland, 1998).
Google Scholar 
Lieberman, D. E. The Story of the Human Body: Evolution, Health, and Disease (Pantheon Books, New York, 2013).
Google Scholar 
Braun, D. et al. Early hominin diet included diverse terrestrial and aquatic animals 1.95 Ma in East Turkana, Kenya. Proc. Natl. Acad. Sci. USA 107, 10002–7 (2010). https://doi.org/10.1073/pnas.1002181107.O’Connor, S., Louys, J., Kealy, S. & Samper Carro, S. C. Hominin dispersal and settlement east of huxley’s line: The role of sea level changes, island size, and subsistence behavior. Curr. Anthropol. 58(S17), S567–S582 (2017).Macaulay, V. et al. Single, rapid coastal settlement of asia revealed by analysis of complete mitochondrial genomes. Science (New York, N.Y.)308, 1034–6 (2005). https://doi.org/10.1126/science.1109792. More

Lack, D. J. Anim. Ecol. 33, 159–173 (1964).Article 

Google Scholar 
Pemberton, J. et al. The unusual value of long-term studies of individuals: the example of the Isle of Rum red deer project. Annu. Rev. Ecol. Evol. Syst. (in the press).Clutton-Brock, T. & Sheldon, B. C. Trends Ecol. Evol. 25, 562–573 (2010).Article 
PubMed 

Google Scholar 
Weimerskirch, H. J. Anim. Ecol. 87, 945–955 (2018).Article 
PubMed 
PubMed Central 

Google Scholar 
Höner, O. P. et al. Nature 448, 798–801 (2007).Article 
PubMed 

Google Scholar 
Rodríguez-Muñoz, R. et al. Evolution 73, 317–328 (2019).Article 
PubMed 
PubMed Central 

Google Scholar 
Czorlich, Y., Aykanat, T., Erkinaro, J., Orell, P. & Primmer, C. R. Science 376, 420–423 (2022).Article 
CAS 
PubMed 

Google Scholar 
Sparkman, A. M., Arnold, S. J. & Bronikowski, A. M. Proc. R. Soc. Lond. B 274, 943–950 (2007).
Google Scholar 
Doak, D. F. & Morris, W. F. Nature 467, 959–962 (2010).Article 
CAS 
PubMed 

Google Scholar 
Campos, F. A. et al. Proc. Natl Acad. Sci. USA 119, e2117669119 (2022).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Forchhammer, M. C., Clutton-Brock, T. H., Lindström, J. & Albon, S. D. J. Anim. Ecol. 70, 721–729 (2001).Article 

Google Scholar 
Bonnet, T. et al. PLoS Biol. 17, e3000493 (2019).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
McCleery, R. H. & Perrins, C. M. Nature 391, 30–31 (1998).Article 
CAS 

Google Scholar 
Charmantier, A. et al. Science 320, 800–803 (2008).Article 
CAS 
PubMed 

Google Scholar 
Vedder, O., Bouwhuis, S. & Sheldon, B. C. PLoS Biol. 11, e1001605 (2013).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Simmonds, E. G., Cole, E. F., Sheldon, B. C. & Coulson, T. Ecol. Lett. 23, 1766–1775 (2020).Article 
PubMed 

Google Scholar 
Cole, E. F., Regan, C. E. & Sheldon, B. C. Nat. Clim. Chang. 11, 872–878 (2021).Article 

Google Scholar 
Huisman, J., Kruuk, L. E., Ellis, P. A., Clutton-Brock, T. & Pemberton, J. M. Proc. Natl Acad. Sci. USA 113, 3585–3590 (2016).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Johnston, S. E., Bérénos, C., Slate, J. & Pemberton, J. M. Genetics 203, 583–598 (2016).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Stoffel, M. A., Johnston, S. E., Pilkington, J. G. & Pemberton, J. M. Nat. Commun. 12, 2972 (2021).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Grieneisen, L. et al. Science 373, 181–186 (2021).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Björk, J. R. et al. Nat. Ecol. Evol. 6, 955–964 (2022).Article 
PubMed 

Google Scholar 
Lamichhaney, S. et al. Science 352, 470–474 (2016).Article 
CAS 
PubMed 

Google Scholar 
Bosse, M. et al. Science 358, 365–368 (2017).Article 
CAS 
PubMed 

Google Scholar 
De Villemereuil, P. et al. Proc. Natl Acad. Sci. USA 117, 31969–31978 (2020).Article 
PubMed 
PubMed Central 

Google Scholar 
Bailey, L. D. et al. Nat. Commun. 13, 2112 (2022).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Bonnet, T. et al. Science 376, 1012–1016 (2022).Article 
CAS 
PubMed 

Google Scholar 
Culina, A. et al. J. Anim. Ecol. 90, 2147–2160 (2021).Article 
PubMed 

Google Scholar  More

Roemmich, D. & Mcgowan, J. Climatic warming and the decline of zooplankton in the California current. Science 267(5202), 1324–1326 (1995).ADS 
CAS 
PubMed 

Google Scholar 
Biard, T. et al. In situ imaging reveals the biomass of giant protists in the global ocean. Nature 532, 504–507 (2016).ADS 
CAS 
PubMed 

Google Scholar 
Irigoien, X., Huisman, J. & Harris, R. P. Global biodiversity patterns of marine phytoplankton and zooplankton. Nature 429, 863–867 (2004).ADS 
CAS 
PubMed 

Google Scholar 
Ware, D. M. & Thomson, R. E. Bottom-up ecosystem trophic dynamics determine fish production in the Northeast Pacific. Science 308(5726), 1280–1284 (2005).ADS 
CAS 
PubMed 

Google Scholar 
Beaugrand, G., Edwards, M. & Legendre, L. Marine biodiversity, ecosystem functioning, and carbon cycles. Proc. Natl. Acad. Sci. 107, 10120–10124 (2010).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Ewald, W. F. Über Orientierung Lokomotion und Lichtreaktionen einiger Cladoceren und deren Bedeutung für die Theorie der Tropismen. Biol. Zentralblatt 30, 1–16 (1910).
Google Scholar 
Dam, H. G., Roman, M. R. & Youngbluth, M. J. Downward export of respiratory carbon and dissolved nitrogen by diel-migrant mesozooplankton at the JGOFS Bermuda time-series station. Deep Sea Res. Part I Oceanogr. Res. Pap. 42, 1187–1197 (1995).ADS 
CAS 

Google Scholar 
Morales, C. E. Carbon and nitrogen fluxes in the ocean: the contribution by zooplankton migrants to active transport in the North Atlantic during the Joint Global Flux Study. J. Plankton Res. 21, 1799–1808 (1999).
Google Scholar 
Steinberg, D. K., Cope, J. S., Wilson, S. E. & Kobari, T. A comparison of mesopelagic mesozooplankton community structure in the subtropical and subarctic North Pacific Ocean. Deep Sea Res. Part II Top. Stud. Oceanogr. 55(14–15), 1615–1635 (2008).ADS 

Google Scholar 
Brugnano, C., Granata, A., Guglielmo, L. & Zagami, G. Spring diel vertical distribution of copepod abundances and diversity in the open Central Tyrrhenian Sea (Western Mediterranean). J. Mar. Syst. 105, 207–220 (2012).
Google Scholar 
Werner, T. & Buchholz, F. Diel vertical migration behaviour in Euphausiids of the northern Benguela current: seasonal adaptations to food availability and strong gradients of temperature and oxygen. J. Plankton Res. 35(4), 792–812 (2013).CAS 

Google Scholar 
Palmer, M. R. & Pearson, P. N. A 23,000-year record of surface water pH and PCO2 in the western equatorial Pacific Ocean. Science 300(5618), 480–482 (2003).ADS 
CAS 
PubMed 

Google Scholar 
Collins, M. et al. The impact of global warming on the tropical Pacific Ocean and El Nino. Nat. Geosci. 3(6), 391–397 (2010).ADS 
CAS 

Google Scholar 
Hu, D. et al. Pacific western boundary currents and their roles in climate. Nature 522, 299–308 (2015).ADS 
CAS 
PubMed 

Google Scholar 
Epp, D. & Smoot, N. C. Distribution of seamounts in the North Atlantic. Nature 337, 254–257 (1989).ADS 

Google Scholar 
Yesson, C., Clark, M. R., Taylor, M. L. & Rogers, A. D. The global distribution of seamounts based on 30 arc seconds bathymetry data. Deep Sea Res. Part I Oceanogr. Res. Pap. 58(4), 442–453 (2011).ADS 

Google Scholar 
Rogers, A. D. The biology of seamounts: 25 Years on. Adv. Mar. Biol. 79, 137–224 (2018).PubMed 

Google Scholar 
Rowden, A. A., Dower, J. F., Schlacher, T. A., Consalvey, M. & Clark, M. R. Paradigms in seamount ecology: fact, fiction and future. Mar. Ecol. 31, 226–241 (2010).ADS 

Google Scholar 
Wilson, R. R. & Kaufmann, R. S. Seamount biota and biogeography. Geophys. Monogr. Ser. 43, 355–377 (2013).ADS 

Google Scholar 
Clark, M. R., Schlacher, T. A., Rowden, A. A., Stocks, K. I. & Consalvey, M. Science priorities for seamounts: research links to conservation and management. PLoS ONE 7(1), e29232 (2012).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Schlacher, T. A., Rowden, A. A., Dower, J. F. & Consalvey, M. Seamount science scales undersea mountains: new research and outlook. Mar. Ecol. 31, 1–13 (2010).ADS 

Google Scholar 
Stocks, K. I. et al. CenSeam, an international program on seamounts within the census of marine life: achievements and lessons learned. PLoS ONE 7(2), e32031 (2012).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Cascao, I., Domokos, R., Lammers, M. O., Santos, R. S. & Silva, M. A. Seamount effects on the diel vertical migration and spatial structure of micronekton. Prog. Oceanogr. 175, 1–13 (2019).ADS 

Google Scholar 
Denda, A., Stefanowitsch, B. & Christiansen, B. From the epipelagic zone to the abyss: trophic structure at two seamounts in the subtropical and tropical Eastern Atlantic – Part II Benthopelagic fishes. Deep Sea Res I Oceanogr. Res. Pap. 130, 78–92 (2017).ADS 
CAS 

Google Scholar 
Dai, L. et al. Zooplankton abundance, biovolume and size spectra down to 3000 m depth in the western tropical North Pacific during autumn 2014. Deep Sea Res. Part I Oceanogr. Res. Pap. 121, 1–13 (2017).ADS 

Google Scholar 
Sun, D., Zhang, D. S., Zhang, R. Y. & Wang, C. S. Different vertical distribution of zooplankton community between North Pacific Subtropical Gyre and Western Pacific Warm Pool: its implication to carbon flux. Acta Oceanol. Sin. 38(6), 32–45 (2019).CAS 

Google Scholar 
Behrenfeld, M. J. et al. Global satellite-observed daily vertical migrations of ocean animals. Nature 576, 257–261 (2019).CAS 
PubMed 

Google Scholar 
Haury, L., Fey, C., Newland, C. & Genin, A. Zooplankton distribution around four eastern North Pacific seamounts. Prog. Oceanogr. 45(1), 69–105 (2000).ADS 

Google Scholar 
Genin, A. Bio-physical coupling in the formation of zooplankton and fish aggregations over abrupt topographies. J. Mar. Syst. 50(1–2), 3–20 (2004).
Google Scholar 
Valle-Levinson, A., Castro, A. T., de Velasco, G. G. & Armas, R. G. Diurnal vertical motions over a seamount of the southern Gulf of California. J. Mar. Syst. 50(1–2), 61–77 (2004).
Google Scholar 
Martin, B. & Christiansen, B. Distribution of zooplankton biomass at three seamounts in the NE Atlantic. Deep Sea Res. Part II Top. Stud. Oceanogr. 56, 2671–2682 (2009).ADS 
CAS 

Google Scholar 
Rawlinson, K. A., Davenport, J. & Barnes, D. K. A. Vertical migration strategies with respect to advection and stratification in a semi-enclosed lough: a comparison of mero- and holozooplankton. Mar. Biol. 144, 935–946 (2004).
Google Scholar 
Forward, R. B. Diel vertical migration: zooplankton photobiology and behaviour. Oceanogr. Mar. Biol. Ann. Rev. 26, 361–393 (1988).
Google Scholar 
Tao, Z. C., Wang, Y. Q., Wang, J. J., Liu, M. T. & Zhang, W. C. Photobehaviors of the calanoid copepod Calanus sinicus from the Yellow Sea to visible and UV-B radiation as a function of wavelength and intensity. J. Oceanol. Limnol. 37(4), 1289–1300 (2019).ADS 

Google Scholar 
Fragopoulu, N. & Lykakis, J. J. Vertical distribution and nocturnal migration of zooplankton in relation to the development of the seasonal thermocline in Patraikos Gulf. Mar. Biol. 104(3), 381–387 (1990).
Google Scholar 
Lougee, L. A., Bollens, S. M. & Avent, S. R. The effects of haloclines on the vertical distribution and migration of zooplankton. J. Exp. Mar. Biol. Ecol. 278(2), 111–134 (2002).
Google Scholar 
Saltzman, J. & Wishner, K. F. Zooplankton ecology in the eastern tropical Pacific oxygen minimum zone above a seamount: 2. Vertical distribution of copepods. Deep Sea Res. Part I Oceanogr. Res. Pap. 44(6), 931–954 (1997).ADS 
CAS 

Google Scholar 
Antezana, T. Species-specific patterns of diel migration into the oxygen minimum zone by euphausiids in the Humboldt Current Ecosystem. Prog. Oceanogr. 83, 228–236 (2009).ADS 

Google Scholar 
Johnsen, G. H. & Jakobsen, P. J. The effect of food limitation on vertical migration in Daphnia longispina. Limnol. Oceanogr. 32(4), 873–880 (1987).ADS 

Google Scholar 
Spinelli, M. et al. Diel vertical distribution of the larvacean Oikopleura dioica in a North Patagonian tidal frontal system (42 degrees-45 degrees S) of the SW Atlantic Ocean. Mar. Biol. Res. 11(6), 633–643 (2015).
Google Scholar 
Guillam, M. et al. Vertical distribution of brittle star larvae in two contrasting coastal embayments: implications for larval transport. Sci. Rep. 10(1), 1–5 (2020).
Google Scholar 
Stramma, L. et al. Expansion of oxygen minimum zones may reduce available habitat for tropical pelagic fishes. Nat. Clim. Change 2(1), 33–37 (2012).ADS 
CAS 

Google Scholar 
Ma, J. et al. The OMZ and its influence on POC in the Tropical Western Pacific Ocean: based on the survey in March 2018. Front. Earth Sci. 9, 632229 (2021).
Google Scholar 
Sun, Q. Q., Song, J. M., Li, X. G., Yuan, H. M. & Wang, Q. D. The bacterial diversity and community composition altered in the oxygen minimum zone of the Tropical Western Pacific Ocean. J. Oceanol. Limnol. 39(5), 1690–1704 (2021).ADS 
CAS 

Google Scholar 
Wang, Q. D. et al. Characteristics and biogeochemical effects of oxygen minimum zones in typical seamount areas, Tropical Western Pacific. J. Oceanol. Limnol. 39(5), 1651–1661 (2021).ADS 
CAS 

Google Scholar 
Fernández-Álamo, M. A. & Färber-Lorda, J. Zooplankton and the oceanography of the eastern tropical Pacific: a review. Prog. Oceanogr. 69(2–4), 318–359 (2006).ADS 

Google Scholar 
Wishner, K. F., Gowing, M. M. & Gelfman, C. Living in suboxia: Ecology of an Arabian Sea oxygen minimum zone copepod. Limnol. Oceanogr. 45(7), 1576–1593 (2000).ADS 

Google Scholar 
Wishner, K. F. et al. Vertical zonation and distributions of calanoid copepods through the lower oxycline of the Arabian Sea oxygen minimum zone. Prog. Oceanogr. 78(2), 163–191 (2008).ADS 

Google Scholar 
Ekau, W., Auel, H., Portner, H. O. & Gilbert, D. Impacts of hypoxia on the structure and processes in pelagic communities (zooplankton, macro-invertebrates and fish). Biogeosciences 7(5), 1669–1699 (2010).ADS 
CAS 

Google Scholar 
Hernández-León, S. et al. Zooplankton and micronekton active flux across the tropical and subtropical Atlantic Ocean. Front. Mar. Sci. 6, 535 (2019).
Google Scholar 
Steinberg, D. K. & Landry, M. R. Zooplankton and the ocean carbon cycle. Ann. Rev. Mar. Sci. 9, 413–444 (2017).PubMed 

Google Scholar 
Le Borgne, R. & Rodier, M. Net zooplankton and the biological pump: a comparison between the oligotrophic and mesotrophic equatorial Pacific. Deep Sea Res. Part II Top. Stud. Oceanogr. 44, 2003–2023 (1997).ADS 

Google Scholar 
Al-Mutairi, H. & Landry, M. R. Active export of carbon and nitrogen at Station ALOHA by diel migrant zooplankton. Deep Sea Res. Part II Top. Stud. Oceanogr. 48, 2083–2103 (2001).ADS 
CAS 

Google Scholar 
Ge, R., Chen, H., Zhuang, Y. & Liu, G. Active carbon flux of mesozooplankton in South China Sea and Western Philippine Sea. Front. Mar. Sci. 8, 1324 (2021).
Google Scholar 
Steinberg, D. K. et al. Bacterial vs. zooplankton control of sinking particle flux in the ocean’s twilight zone. Limnol. Oceanogr. 53(4), 1327–1338 (2008).ADS 

Google Scholar 
Hirch, S., Martin, B. & Christiansen, B. Zooplankton metabolism and carbon demand at two seamounts in the NE Atlantic. Deep Sea Res. Part II Top. Stud. Oceanogr. 56(25), 2656–2670 (2009).ADS 
CAS 

Google Scholar 
Denda, A. & Christiansen, B. Zooplankton distribution patterns at two seamounts in the subtropical and tropical NE Atlantic. Mar. Ecol. 35(2), 159–179 (2014).ADS 

Google Scholar 
Dower, J. F. & Mackas, D. L. “Seamount effects” in the zooplankton community near Cobb Seamount. Deep Sea Res. Part I Oceanogr. Res. Pap. 43, 837–858 (1996).ADS 

Google Scholar 
Ma, J. et al. Analysis of differences in nutrients chemistry in seamount seawaters in the Kocebu and M5 seamounts in Western Pacific Ocean. J. Oceanol. Limnol. 39(5), 1662–1674 (2021).ADS 

Google Scholar 
Denda, A., Mohn, C., Wehrmann, H. & Christiansen, B. Microzooplankton and meroplanktonic larvae at two seamounts in the subtropical and tropical NE Atlantic. J. Mar. Biol. Assoc. U. K. 97(1), 1–27 (2017).
Google Scholar 
Tutasi, P. & Escribano, R. Zooplankton diel vertical migration and downward C flux into the oxygen minimum zone in the highly productive upwelling region off northern Chile. Biogeosciences 17(2), 455–473 (2020).ADS 
CAS 

Google Scholar 
Harris, R., Wiebe, P., Lenz, J., Skjoldal, H. R. & Huntley, M. ICES Zooplankton Methodology Manual (Academic Press, 2000).
Google Scholar 
Zhang, X. & Dam, H. G. Downward export of carbon by diel migrant mesozooplankton in the central equatorial Pacific. Deep Sea Res. Part II Top. Stud. Oceanogr. 44, 2191–2202 (1997).ADS 
CAS 

Google Scholar 
Isla, A., Scharek, R. & Latasa, M. Zooplankton diel vertical migration and contribution to deep active carbon flux in the NW Mediterranean. J. Mar. Syst. 143, 86–97 (2015).
Google Scholar 
Ikeda, T. Respiration and ammonia excretion by marine metazooplankton taxa: synthesis toward a global-bathymetric model. Mar. Biol. 161(12), 2753–2766 (2014).CAS 

Google Scholar 
Steinberg, D. K. et al. Zooplankton vertical migration and the active transport of dissolved organic and inorganic carbon in the Sargasso Sea. Deep Sea Res. Part I Oceanogr. Res. Pap. 47(1), 137–158 (2000).ADS 
CAS 

Google Scholar 
Andersen, V. et al. Vertical distributions of zooplankton across the Almeria-Oran frontal zone (Mediterranean Sea). J. Plankton Res. 26(3), 275–293 (2004).
Google Scholar  More

Study sites and data collectionDuring 2017 and 2018, we carried out 14 experiments in rivers located in temperate, tropical, and subarctic biomes to capture a gradient of river productivity and climatic characteristics (Table 1, Fig. 1). Apart from the Mekong and Sekong rivers in Cambodia that were impacted by plantations, rice cultivation, grassland, and urban areas (56% impacted land cover in the Mekong and 38% in the Sekong), the selected rivers were predominantly in pristine areas (impacted land-use ≤ 8%), although two rivers in Mongolia were affected by livestock grazing (with 26% of land cover at the Khovd and 59% in the two Zavkhan rivers).We conducted traditional O2 concentration metabolic assessments, assessments of isotopic fractionation, and 24 h characterization of δ18O2 at each site. We measured changes in dissolved O2 concentrations and temperature every 10 min over at least 24 h with at least one MiniDOT logger (PME, Vista, California, USA). We calibrated for drift using the average measurement values made in 100% saturated water for at least 30 min before and after each deployment to allow adjustment to temperature and placed sensors in the river for at least 30 min prior to using data to allow equilibration to temperature (following methods detailed in ref. 52).We collected δ18O2 samples by hand every 2 h during the same 24-h period of the O2 concentration measurements in pre-evacuated 100 mL vials loaded with 50 µl HgCl2 as a preservative and sealed with septum stoppers (Bellco Glass Inc., Supelco, Vineland NJ). We analyzed samples for δ18O2 at the Nevada Stable Isotope Lab of the University of Nevada, Reno with a Micromass Isoprime (Middlewich, UK) stable isotope ratio mass spectrometer. We followed the method described by ref. 17 and injected 1.0–2.5 mL of headspace gas taken from the serum bottles using a gastight syringe (SGE, Australia) into a Eurovector (Pavia, Italy) elemental analyzer equipped with a septum injector port, and a 1.5 m long molecular sieve gas chromatography column. Water-δ18O was also collected at each site every 2 h and analyses were performed using a Picarro L2130-i cavity ringdown spectrometer at the Nevada Stable Isotope Lab of the University of Nevada, Reno. δ18O2 values are reported in the usual δ notation vs. VSMOW in units of ‰, with an analytical uncertainty of ±0.2‰ for δ18O2, or an analytical uncertainty of ±0.1‰ for water-δ18O.We characterized physical characteristics at each site to provide parameters to estimate whole-system metabolism. We measured conductivity, slope, and flow velocity and depth at ten transects using a flow meter when wadeable or with an Acoustic Doppler Velocimeter (Sontek, Xylem, San Diego, CA) when rivers were not wadeable. At each site, we measured light as photosynthetically active radiation (PAR) every 10 min, using Odyssey PAR loggers (Data Flow Systems, Christchurch, New Zealand) calibrated with a Li-Cor PAR sensor (Lincoln, Nebraska, USA).At each site, we also directly measured biofilm ash-free dry mass (AFDM) from 8 to 12 rocks (53). The material was scrubbed from the rocks, agitated, filtered (Whatman glass microfiber GF/F filters). Rock area was estimated with calibrated pictures processed with the ImageJ processing program (National Institutes of Health and the Laboratory for Optical and Computational Instrumentation LOCI, University of Wisconsin). For AFDM analyses, samples were dried, and weighed before and after combustion.Additionally, we collected data on the percentage of impacted land use in the watershed above each sampling site: for the Mekong and the Sekong we used Landsat satellite imagery from ref. 54, for the US and Mongolian sites land use characteristics were derived from the National Land Cover Database55 and for Patagonia we used the Chilean national land use inventory maps from ref. 56.δ18O2 stable isotope fractionation during respiration in sealed recirculating chambersModels based on oxygen isotopes are sensitive to the oxygen isotope fractionation factor (αR) during respiration used; αR can vary widely among sites and is influenced by temperature and water velocity30. We used in our models the range of αR values measured by30 using sealed Plexiglas recirculating chambers as in ref. 57. These measurements were done at the same time as the 24 h δ18O2 sample collections in the rivers of this study. We placed rocks, sediment, macrophytes (macrophytes dominated in the Zavkhan 1 site) inside the chambers, depending on the site’s dominant substrata (see ref. 30 for more details on chamber measurements). We collected water samples in the chambers for δ18O2 analyses before and after the incubations and the O2 isotope fractionation factor was calculated using Eq. (2).$$delta =(delta i+1000){F}^{left(alpha -1right)}-1000$$
(2)
where δ is the O2 isotopic composition of dissolved oxygen at the end of the dark incubation, δi is the O2 isotopic composition of dissolved oxygen at the beginning of the dark incubation, F the fractional abundance of O2 concentration remaining at the end of the dark incubation, and α is the isotopic fractionation factor during respiration.Ecosystem metabolism O2 single station modelingWe modeled metabolism as a function of GPP, ER, and reaeration with the atmosphere, using the single-station open-channel metabolism method4 using the same approach as15, given in Eq. (3).$${O}_{{2}_{(t)}}={O}_{{2}_{(t-1)}}+left(left(frac{{GPP}}{z}xfrac{{{PPFD}}_{left(t-1right)}}{sum {{PPFD}}_{24h}}right)+frac{{ER}}{z}+{K}_{{O}_{2}}left({O}_{{2}_{{sat}left(t-1right)}}-{O}_{{2}_{left(t-1right)}}right)right)triangle t$$
(3)
where GPP is gross primary production in g O2 m−2 d−1, ER is ecosystem respiration in g O2 m−2 d−1, ({K}_{{O}_{2}}) is the reaeration coefficient (d−1). PPFD is photosynthetic photon flux density (µmol m−2 s−1), z is mean stream depth (m), and ∆t is time increment between logging intervals (d). We used Bayesian inverse modeling approach to estimate the probability distribution of parameters GPP and ER that produce the best model fit between observed and modeled O2 data. We fixed site-specific ({K}_{{O}_{2}}) estimates using K600 (d−1) (normalized beyond gas-specific Schmidt number conversions among gases58) based on prior work characterizing K using BASE59, and converted these prior estimates of K600 to ({K}_{{O}_{2}})using appropriate temperature corrections. We estimated daily GPP and ER from diel O2 data only (Eq. (3)) to be used as prior estimates of daily GPPO2 and ERO2 in the coupled O2 and δ18O2 model (Eqs. (4a) and (4b))15, where the mean and SD of GPP and ER from the O2 _only method were used as prior estimates of GPPO2 and ERO2 in the dual O2 and δ18O2 model described below.Ecosystem metabolism: Diel δ18O2 modelingWe also modeled metabolism using an updated version of the model developed by ref. 15 coupling high-frequency O2 concentration data with δ18O2 collected every 2 h throughout the same 24 h period of the O2 concentration measurements. With this model, daily rates of ecosystem metabolism are derived from diel changes in δ18O2 and O2, where values of δ18O2 are converted to g 18O m−3 (18O2 in Eq. 4b) and modeled as a function of water isotope values, isotope fractionation, reaeration with the atmosphere, ER, and GPP. As with Eq. 3, the ratio of light at the previous logging time (({{PPFD}}_{left(t-1right)})) relative to the sum of light over 24 h (({sum {PPFD}}_{24h})) is used to characterize times when GPP is zero and only ER is taking place (Eqs. (4a) and (4b)):$${O}_{{2}_{left(tright)}}= , {O}_{{2}_{left(t-1right)}}+left(frac{{{GPP}}_{O2}}{z}xfrac{{{PPFD}}_{left(t-1right)}}{sum {{PPFD}}_{24h}}right)+left(frac{{{ER}}_{O2},xtriangle t}{z}right)\ +left({K}_{{O}_{2}}xleft({O}_{{2}_{{sat}left(t-1right)}}-{O}_{{2}_{left(t-1right)}}right)xtriangle tright)$$
(4a)
$${18O}_{{2}_{(t)}}=, {18O}_{{2}_{(t-1)}}+left(frac{left({{GPP}}_{O2}+{dielMET}right)}{z}xfrac{{{PPFD}}_{left(t-1right)}}{{sum {PPFD}}_{24h}}x,{alpha }_{P},x,{{AF}}_{W}right)\ +left(frac{{{ER}}_{O2},xtriangle t}{z}x,{alpha }_{R},x,{{AF}}_{{DO}}left(t-1right)right)\ +left(frac{left(-{dielMET}right)}{z}xfrac{{{PPFD}}_{left(t-1right)}}{sum {{PPFD}}_{24h}}x,{alpha }_{R},x,{{AF}}_{{DO}}left(t-1right)right)\ +left({K}_{{O}_{2}}x,{alpha }_{g}xtriangle t,xleft(left({O}_{{2}_{{sat}left(t-1right)}}x,{alpha }_{g},x,{{AF}}_{{atm}}right)-{18O}_{{2}_{(t-1)}}right)right)$$
(4b)
Where GPPO2 and ERO2 (g O2 m−2 d−1) refer to the values obtained from diel O2 only, dielMET (g O2 m−2 d−1) is the diel metabolism term that allows for the estimation of diel ER and GPP from 18O2, KO2 is the O2 gas exchange rate (d−1), z is mean stream depth (m), PPFD is photosynthetic photon flux density (µmol m−2 s−1), Δt is time step between measurements (d), 18O2 is the concentration of 18O in dissolved O2 (g 18O m−3), AFDO is atomic fraction of dissolved O2 (mol18O:mol O2, measured), AFw is atomic fraction of H2O (mol 18O:mol O2, measured), AFatm is atomic fraction of atmospheric air (mol18O:mol O2, literature), αg is the fractionation factor during air–water gas exchange (0.9972, from ref. 60), αR is the fractionation factor during respiration measured in the chambers (varied by site30; Fig. 1), αp is the fractionation factor during photosynthesis (1.0000 from ref. 60).The inverse modeling approach finds the best estimates of parameters to match measured and modeled dissolved O2. The model assumes that the measured changes in O2 concentration represent the actual net diel changes in O2 concentration and uses an additional parameter, dielMET, that is a function of the isotopic enrichment occurring during respiration, derived from diel 18O2. This parameter increases daily ERO2 and GPPO2 of the same amount, adding and subtracting dielMET, to obtain daily δ18O2-ER and δ18O2-GPP, respectively.We estimated the posterior distributions of unknown parameters (ERO2, GPPO2, and dielMET) using a Bayesian inverse modeling approach15 and Markov chain Monte Carlo sampling with the R metrop function in the mcmc package61,62. Each model was run for at least 200,000 iterations using nominally informative priors based on the range of ERO2 and GPPO2. For dielMET, we used a minimally informative uniform prior distribution (0–100 g O2 m−2 d−1). We removed the first 10,000 iterations of model burn-in and assessed quality of model fit. Model runs using the minimum, average, and maximum αR values measured in the field recirculating chambers were also compared, and we selected the αR and report associated model metabolism estimates that generated the lowest sum of squared differences between the observed and modeled O2 and 18O2 diel values.Temperature-normalized comparisonsTo test the effect of temperature from the daily δ18O2-ER and δ18O2-GPP rates and account for daily variations in temperature, we normalized estimates from models to 20 °C (and report them as 20δ18O2-ER and 20δ18O2-GPP) for comparison with O2-derived metabolism estimates following33 with Eq. (5):$${rate},{at},20,{}^circ C=frac{{2.523* e}^{(0.0552* 20)}}{{2.523* e}^{(0.0552* {t}_{1})},* {rate},{at},{t}_{1}}$$
(5)
Where t1 is site temperature and rate is the measured rate (i.e., GPP or ER) at t1.Statistical analysesWe used multiple linear regression to find the best predictor of the magnitude of diel 20δ18O2-ER and differences between sites. To select the best model, we performed a stepwise variable selection and selected the best model based on the lowest AIC. Tested variables included percentage of impacted land use (%), 20δ18O2-GPP (g O2 m−2 d−1), conductivity (µS/cm), ash-free dry mass (AFDM, g), slope (%), water depth (m), and flow velocity (m/s) measured in the field. We used ANOVA to test the relative contribution of each variable selected with the AIC to total variance. Analyses were run with the R software61.Reporting summaryFurther information on research design is available in the Nature Portfolio Reporting Summary linked to this article. More