Philippa Kaur

Schmidt, J. Breeding places and migrations of the eel. Nature 111, 51–54 (1923).ADS 
Article 

Google Scholar 
Tucker, D. W. A new solution to the Atlantic eel problem. Nature 183, 495–501 (1959).ADS 
Article 

Google Scholar 
Voorhis, A. D. & Hersey, J. B. Oceanic thermal fronts in the Sargasso Sea. J. Geophys. Res. 69(18), 3809–3814 (1964).ADS 
Article 

Google Scholar 
Kleckner, R. C. & McCleave, J. D. The northern limit of spawning by Atlantic eels (Anguilla spp.) in the Sargasso Sea in relation to thermal fronts and surface water masses. J. Mar. Res. 46, 647–667 (1988).Article 

Google Scholar 
Ullman, D. S., Cornillon, P. C. & Shan, Z. On the characteristics of subtropical fronts in the North Atlantic. J. Geophys. Res: Oceans 112, C01010 (2007).ADS 

Google Scholar 
Miller, M. J. et al. Spawning by the European eel across 2000 km of the Sargasso Sea. Biol. Lett. 15, 20180835 (2019).Article 

Google Scholar 
Westerberg, H. et al. Larval abundance across the European eel spawning area: An analysis of recent and historic data. Fish. 19, 890–902 (2018).
Google Scholar 
Halliwell, G. R. Jr., Olson, D. B. & Peng, G. Stability of the Sargasso Sea subtropical frontal zone. J. Phys. Oceanogr. 24(6), 1166–1183 (1994).ADS 
Article 

Google Scholar 
van Ginneken, V. J. T. & Maes, G. E. The European eel (Anguilla anguilla, Linnaeus), its lifecycle, evolution and reproduction: A literature review. Rev. Fish Biol. Fish. 15, 367–398 (2005).Article 

Google Scholar 
Friedland, K. D., Miller, M. J. & Knights, B. Oceanic changes in the Sargasso Sea and declines in recruitment of the European eel. ICES J. Mar. Sci. 64, 519–530 (2007).Article 

Google Scholar 
Jacoby, D. M. P. et al. Synergistic patterns of threat and the challenges facing global anguillid eel conservation. Glob. Ecol. Conserv. 4, 321–333 (2015).Article 

Google Scholar 
Béguer-Pon, M. et al. Tracking anguillid eels: Five decades of telemetry-based research. Mar. Freshw. Res. 69, 199 (2018).Article 

Google Scholar 
Righton, D. et al. Important questions to progress science and sustainable management of anguillid eels. Fish 22, 762–788 (2021).
Google Scholar 
Aoyama, J. Life history and evolution of migration in catadromous eels (genus Anguilla). Aquat. Bio Sci. Monogr. 2, 1–42 (2009).
Google Scholar 
Tsukamoto, K., Aoyama, J. & Miller, M. J. Migration, speciation, and the evolution of diadromy in anguillid eels. Can. J. Fish. Aquat. Sci. 59, 1989–1998 (2002).Article 

Google Scholar 
Tesch, F.-W. Telemetric observations on the spawning migration of the eel (Anguilla anguilla) west of the European continental shelf. Env. Biol. Fish. 3, 203–209 (1978).Article 

Google Scholar 
Aarestrup, K. et al. Oceanic spawning migration of the European eel (Anguilla anguilla). Science 325, 1660 (2009).ADS 
CAS 
Article 

Google Scholar 
Westerberg, H. et al. Behaviour of stocked and naturally recruited European eels during migration. Mar. Ecol. Prog. Ser. 496, 145–157 (2014).ADS 
Article 

Google Scholar 
Amilhat, E. et al. First evidence of European eels exiting the Mediterranean Sea during their spawning migration. Sci. Rep. 6, 21817 (2016).ADS 
CAS 
Article 

Google Scholar 
Righton, D. et al. Empirical observations of the spawning migration of European eels: The long and dangerous road to the Sargasso Sea. Sci. Adv. 2, e1501694 (2016).ADS 
Article 

Google Scholar 
Verhelst, P. et al. Mapping silver eel migration routes in the North Sea. Sci Rep. 12, 318 (2022).ADS 
CAS 
Article 

Google Scholar 
Kuroki, M. et al. Hatching time and larval growth of Atlantic eels in the Sargasso Sea. Mar. Biol. 164, 118. https://doi.org/10.1007/s00227-017-3150-9 (2017).Article 

Google Scholar 
Acton, L. et al. What is the Sargasso Sea? The problem of fixing space in a fluid ocean. Polit. Geogr. 68, 86–100 (2019).Article 

Google Scholar 
GEBCO Compilation Group. GEBCO 2020 Grid. https://doi.org/10.5285/a29c5465-b138-234d-e053-6c86abc040b9 (2020).Miller, M. J. & Hanel, R. The Sargasso Sea Subtropical Gyre: The spawning and larval development area of both freshwater and marine eels. Sargasso Sea Alliance Science Report Series, 7, 20 pp (2011).Munk, P. et al. Oceanic fronts in the Sargasso Sea control the early life and drift of Atlantic eels. Proc. Biol. Sci. 277, 3593–3599 (2010).PubMed 
PubMed Central 

Google Scholar 
Béguer-Pon, M., Castonguay, M., Shan, S., Benchetrit, J. & Dodson, J. J. Direct observations of American eels migrating across the continental shelf to the Sargasso Sea. Nat. Commun. 6, 8705 (2015).ADS 
Article 

Google Scholar 
Westin, L. The spawning migration of European silver eel (Anguilla anguilla L.) with particular reference to stocked eel in the Baltic. Fish. Res. 38(3), 257–270 (1998).
Article 

Google Scholar 
Tesch, F.-W., Wendt, T. & Karlsson, L. Influence of geomagnetism on the activity and orientation of the eel, Anguilla anguilla (L.), as evident from laboratory experiments. Ecol. Freshw. Fish 1(1), 52–60 (1992).Article 

Google Scholar 
Tesch, F.-W. The Eel (Blackwell Science, Oxford, UK, 2003).Book 

Google Scholar 
Durif, C. M. F. et al. Magnetic compass orientation in the European eel. PLoS ONE 8(3), e59212 (2013).ADS 
CAS 
Article 

Google Scholar 
Schabetsberger, R. et al. Hydrographic features of anguillid spawning areas: Potential signposts for migrating eels. Mar. Ecol. Prog. Ser. 554, 141–155 (2016).ADS 
Article 

Google Scholar 
Naisbett-Jones, L. C., Putman, N. F., Stephenson, J. F., Ladak, S. & Young, K. A. A magnetic map leads juvenile European eels to the Gulf stream. Curr. Biol. 27, 1236–1240 (2017).CAS 
Article 

Google Scholar 
Dekker, W. Status of the European eel stock and fisheries. In Eel Biology (eds Aida, K. et al.) 237–254 (Springer, New York, 2003).Chapter 

Google Scholar 
Drouineau, H. et al. Freshwater eels: A symbol of the effects of global change. Fish Fish (Oxf) 19, 903–930 (2018).Article 

Google Scholar 
ICES. Joint EIFAAC/ICES/GFCM Working Group on Eels (WGEEL). ICES Scientific Reports. 2(85) (2020).Pike, C., Crook, V. & Gollock, M. Anguilla anguilla. The IUCN Red List of Threatened Species 2020: e.T60344A152845178 (2020).Durif, C., Dufour, S. & Elie, P. The silvering process of Anguilla anguilla: A new classification from the yellow resident to the silver migrating stage. J. Fish. Biol. 66, 1025–1043 (2005).Article 

Google Scholar 
Pankhurst, N. W. Relation of visual changes to the onset of sexual maturation in the European eel Anguilla Anguilla (L.). J. Fish Biol. 21, 127–140 (1982).Article 

Google Scholar 
Økland, F., Thorstad, E. B., Westerberg, H., Aarestrup, K. & Metcalfe, J. D. Development and testing of attachment methods for pop-up satellite archival transmitters in European eel. Anim. Biotelem. 1, 3 (2013).Article 

Google Scholar  More

Vega, F. E., Rosenquist, E. & Collins, W. Global project needed to tackle coffee crisis. Nature 425, 343 (2003).ADS 
CAS 
PubMed 
Article 

Google Scholar 
Craparo, A. C. W., Van Asten, P. J. A., Läderach, P., Jassogne, L. T. P. & Grab, S. W. Coffea arabica yields decline in Tanzania due to climate change: global implications. Agric. For. Meteorol. 207, 1–10 (2015).ADS 
Article 

Google Scholar 
Davis, A. P. et al. High extinction risk for wild coffee species and implications for coffee sector sustainability. Sci. Adv. 5, eaav3473 (2019).ADS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Davis, A. P., Gole, T. W., Baena, S. & Moat, J. The impact of climate change on indigenous arabica coffee (Coffea arabica): predicting future trends and identifying priorities. PLoS ONE 7, e47981 (2012).ADS 
CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Davis, A. P., Mieulet, D., Moat, J., Sarmu, D. & Haggar, J. Arabica-like flavour in a heat-tolerant wild coffee species. Nat. Plants 7, 413–418 (2021).CAS 
PubMed 
Article 

Google Scholar 
Moat, J., Gole, T. W. & Davis, A. P. Least concern to endangered: applying climate change projections profoundly influences the extinction risk assessment for wild Arabica coffee. Global Change Biol. 25, 390–403 (2019).ADS 
Article 

Google Scholar 
Moat, J. et al. Resilience potential of the Ethiopian coffee sector under climate change. Nat. Plants 3, 17081 (2017).PubMed 
Article 

Google Scholar 
Kath, J. et al. Not so robust: Robusta coffee production is highly sensitive to temperature. Global Change Biol. 26, 3677–3688 (2020).ADS 
Article 

Google Scholar 
Liu, L. et al. Soil moisture dominates dryness stress on ecosystem production globally. Nat. Commun. 11, 1–9 (2020).ADS 
CAS 

Google Scholar 
Grossiord, C. et al. Plant responses to rising vapor pressure deficit. New Phytol. 226, 1550–1566 (2020).PubMed 
Article 

Google Scholar 
IPCC Climate Change 2022: Impacts, Adaptation, and Vulnerability (eds. Pörtner, H.-O. et al.) (Cambridge Univ. Press, 2022).Burke, M. et al. Higher temperatures increase suicide rates in the United States and Mexico. Nat. Clim. Change 8, 723–729 (2018).ADS 
Article 

Google Scholar 
Burke, M., Hsiang, S. M. & Miguel, E. Global non-linear effect of temperature on economic production. Nature 527, 235–239 (2015).ADS 
CAS 
PubMed 
Article 

Google Scholar 
Duffy, K. A. et al. How close are we to the temperature tipping point of the terrestrial biosphere? Sci. Adv. 7, eaay1052 (2021).ADS 
CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Scheffer, M. et al. Early-warning signals for critical transitions. Nature 461, 53–59 (2009).ADS 
CAS 
PubMed 
Article 

Google Scholar 
Schneider, S. H. Abrupt non-linear climate change, irreversibility and surprise. Global Environ. Change 14, 245–258 (2004).Article 

Google Scholar 
Lenton, T. M. Early warning of climate tipping points. Nat. Clim. Change 1, 201–209 (2011).ADS 
Article 

Google Scholar 
Lenton, T. M. et al. Climate tipping points—too risky to bet against. Nature. 575, 592–595 (2019).ADS 
CAS 
PubMed 
Article 

Google Scholar 
Lobell, D. B., Bänziger, M., Magorokosho, C. & Vivek, B. Nonlinear heat effects on African maize as evidenced by historical yield trials. Nat. Clim. Change 1, 42–45 (2011).ADS 
Article 

Google Scholar 
Lobell, D. B., Deines, J. M. & Tommaso, S. D. Changes in the drought sensitivity of US maize yields. Nat. Food 1, 729–735 (2020).Article 

Google Scholar 
Lobell, D. B. et al. Greater sensitivity to drought accompanies maize yield increase in the US Midwest. Science 344, 516–519 (2014).ADS 
CAS 
PubMed 
Article 

Google Scholar 
Rigden, A., Mueller, N., Holbrook, N., Pillai, N. & Huybers, P. Combined influence of soil moisture and atmospheric evaporative demand is important for accurately predicting US maize yields. Nat. Food 1, 127–133 (2020).Article 

Google Scholar 
Schlenker, W. & Roberts, M. J. Nonlinear temperature effects indicate severe damages to US crop yields under climate change. Proc. Natl Acad. Sci. USA 106, 15594–15598 (2009).ADS 
CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
McDowell, N. G. et al. Mechanisms of woody-plant mortality under rising drought, CO2 and vapour pressure deficit. Nat. Rev. Earth Environ. 3, 294–308 (2022).ADS 
CAS 
Article 

Google Scholar 
Sinclair, T. R. et al. Limited-transpiration response to high vapor pressure deficit in crop species. Plant Sci. 260, 109–118 (2017).CAS 
PubMed 
Article 

Google Scholar 
López, J., Way, D. A. & Sadok, W. Systemic effects of rising atmospheric vapor pressure deficit on plant physiology and productivity. Global Change Biol. 27, 1704–1720 (2021).ADS 
Article 

Google Scholar 
McDowell, N. G. & Allen, C. D. Darcy’s law predicts widespread forest mortality under climate warming. Nat. Clim. Change 5, 669–672 (2015).ADS 
Article 

Google Scholar 
Abatzoglou, J. T., Dobrowski, S. Z., Parks, S. A. & Hegewisch, K. C. TerraClimate, a high-resolution global dataset of monthly climate and climatic water balance from 1958–2015. Sci. Data 5, 170191 (2018).PubMed 
PubMed Central 
Article 

Google Scholar 
You, L., Wood, S., Wood-Sichra, U. & Wu, W. Generating global crop distribution maps: from census to grid. Agric. Syst. 127, 53–60 (2014).Article 

Google Scholar 
Fong, Y., Huang, Y., Gilbert, P. B. & Permar, S. R. chngpt: threshold regression model estimation and inference. BMC Bioinformatics 18, 1–7 (2017).Article 

Google Scholar 
Qin, Y. et al. Agricultural risks from changing snowmelt. Nat. Clim. Change 10, 459–465 (2020).ADS 
Article 

Google Scholar 
Forster, P. M., Maycock, A. C., McKenna, C. M. & Smith, C. J. Latest climate models confirm need for urgent mitigation. Nat. Clim. Change 10, 7–10 (2020).ADS 
Article 

Google Scholar 
Forster, P. M. et al. Projections of when temperature change will exceed 2 °C above pre-industrial levels. Nat. Clim. Change 10, 407–412 (2011).
Google Scholar 
Joshi, M., Hawkins, E., Sutton, R., Lowe, J. & Frame, D. Projections of when temperature change will exceed 2 °C above pre-industrial levels. Nat. Clim. Change 1, 407–412 (2011).ADS 
Article 

Google Scholar 
IPCC, 2021: Summary for Policymakers. In Climate Change 2021: The Physical Science Basis (eds Masson-Delmotte, V. et al.) (Cambridge Univ. Press, in press).Lobell, D. B. et al. The critical role of extreme heat for maize production in the United States. Nat. Clim. Change 3, 497–501 (2013).ADS 

Google Scholar 
Sinclair, T. R., Hammer, G. L. & Van Oosterom, E. J. Potential yield and water-use efficiency benefits in sorghum from limited maximum transpiration rate. Funct. Plant Biol. 32, 945–952 (2005).PubMed 
Article 

Google Scholar 
Martins, M. Q. et al. Protective response mechanisms to heat stress in interaction with high [CO2] conditions in Coffea spp. Front. Plant Sci. 7, 947 (2016).PubMed 
PubMed Central 
Article 

Google Scholar 
Rodrigues, W. P. et al. Long‐term elevated air [CO2] strengthens photosynthetic functioning and mitigates the impact of supra‐optimal temperatures in tropical Coffea arabica and C. canephora species. Global Change Biol. 22, 415–431 (2016).ADS 
Article 

Google Scholar 
Ghini, R. et al. Coffee growth, pest and yield responses to free-air CO2 enrichment. Clim. Change 132, 307–320 (2015).ADS 
Article 

Google Scholar 
Rakocevic, M. et al. The vegetative growth assists to reproductive responses of Arabic coffee trees in a long-term FACE experiment. Plant Growth Regul. 91, 305–316 (2020).CAS 
Article 

Google Scholar 
Hammer, G. L. et al. Designing crops for adaptation to the drought and high‐temperature risks anticipated in future climates. Crop Sci. 60, 605–621 (2020).Article 

Google Scholar 
Gennari, P., Rosero-Moncayo, J. & Tubiello, F. N. The FAO contribution to monitoring SDGs for food and agriculture. Nat. Plants 5, 1196–1197 (2019).PubMed 
Article 

Google Scholar 
Lesk, C., Rowhani, P. & Ramankutty, N. Influence of extreme weather disasters on global crop production. Nature 529, 84–87 (2016).ADS 
CAS 
PubMed 
Article 

Google Scholar 
Ortiz-Bobea, A., Ault, T. R., Carrillo, C. M., Chambers, R. G. & Lobell, D. B. Anthropogenic climate change has slowed global agricultural productivity growth. Nat. Clim. Change 11, 306–312 (2021).ADS 
Article 

Google Scholar 
Davis, A. P. et al. Hot coffee: the identity, climate profiles, agronomy, and beverage characteristics of Coffea racemosa and C. zanguebariae. Front. Sustain. Food Syst. 5, 740137 (2021).Article 

Google Scholar 
Sarmiento-Soler, A. et al. Disentangling effects of altitude and shade cover on coffee fruit dynamics and vegetative growth in smallholder coffee systems. Agric. Ecosyst. Environ. 326, 107786 (2022).CAS 
Article 

Google Scholar 
Wood, S. N. Fast stable restricted maximum likelihood and marginal likelihood estimation of semiparametric generalized linear models. J. R. Stat. Soc. B 73, 3–36 (2011).MathSciNet 
MATH 
Article 

Google Scholar 
Barton, K. MuMIn: multi-model inference. R-Forge http://r-forge.r-project.org/projects/mumin/ (2009).R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing https://www.r-project.org/ (2021).Harrison, X. A. et al. A brief introduction to mixed effects modelling and multi-model inference in ecology. PeerJ 6, e4794 (2018).PubMed 
PubMed Central 
Article 

Google Scholar 
Najafi, E., Devineni, N., Khanbilvardi, R. M. & Kogan, F. Understanding the changes in global crop yields through changes in climate and technology. Earths Future 6, 410–427 (2018).ADS 
CAS 
Article 

Google Scholar 
Ovalle-Rivera, O. et al. Assessing the accuracy and robustness of a process-based model for coffee agroforestry systems in Central America. Agrofor. Syst. 94, 2033–2051 (2020).Article 

Google Scholar 
Varma, S. & Simon, R. Bias in error estimation when using cross-validation for model selection. BMC Bioinformatics 7, 1–8 (2006).Article 

Google Scholar 
Yuan, W. et al. Increased atmospheric vapor pressure deficit reduces global vegetation growth. Sci. Adv. 5, eaax1396 (2019).ADS 
CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Son, H. & Fong, Y. Fast grid search and bootstrap-based inference for continuous two-phase polynomial regression models. Environmetrics 32, e2664 (2021).MathSciNet 
Article 

Google Scholar 
Wintgens, J. N. et al. Coffee: Growing, Processing, Sustainable Production. A Guidebook for Growers, Processors, Traders, and Researchers (Wiley, 2004). More

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

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

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

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

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

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

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

Google Scholar  More

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

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

Johnson, N. et al. (eds) Global Soil Biodiversity Atlas (EU, 2016).
Google Scholar 
FAO et al. State of Knowledge of Soil Biodiversity — Status, Challenges and Potentialities (FAO, 2020).
Google Scholar 
Cameron, E. K. et al. Nature Ecol. Evol. 2, 1042–1043 (2018).PubMed 
Article 

Google Scholar 
van den Hoogen, J. et al. Nature 572, 194–198 (2019).PubMed 
Article 

Google Scholar 
Phillips, H. R. P. et al. Science 366, 480–485 (2019).PubMed 
Article 

Google Scholar 
Guerra, C. A. et al. Nature https://doi.org/10.1038/s41586-022-05292-x (2022).Article 

Google Scholar 
Moore, J. C. & de Ruiter, P. C. Energetic Food Webs: An Analysis of Real and Model Ecosystems (Oxford Univ. Press, 2012).
Google Scholar 
Wolters V. et al. Bioscience 50, 1089–1098 (2000).Article 

Google Scholar 
Schimel, J. P. & Schaeffer, S. M. Front. Microbiol. 3, 348 (2012).PubMed 
Article 

Google Scholar 
IPCC. In Climate Change 2022: Mitigation of Climate Change. Contribution of Working Group III to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change Impacts, Adaptation, and Vulnerability: Summary for Policymakers (eds Shukla, P. R. et al.) 50 (Cambridge Univ. Press, 2022).
Google Scholar 
Chenu, C. et al. Soil Till. Res. 188, 41–52 (2019).Article 

Google Scholar 
Liang, C., Schimel, J. P. & Jastrow, J. D. Nature Microbiol. 2, 17105 (2017).PubMed 
Article 

Google Scholar 
Hannula, S. E. & Morriën, E. Geoderma 413, 115767 (2022).Article 

Google Scholar  More