Philippa Kaur

If you are a scientist hoping to photograph and share your own research:
•    Don’t underestimate the power of modern media and social-media platforms. Content is changing the world and people’s lives, and it can easily change your life. Stay at the forefront of media technology, or at least be aware of developments. It’s a never-ending race, but it’s easy to get into.
•    If you plan to share your work with others, imagine what will be of interest to them. If you can excitingly describe your work to a 5-year-old, you won’t have any trouble getting anyone interested. Beautiful pictures help, but the story always comes first.

•    You will stand out much more if you have a niche and unique story. It could be your rare field of science or a special angle that you use to tell the story of your work. Being different is awesome.
•    Set the bar very high. You can find dozens of examples of truly high-quality content on the Internet. And you can almost always find resources that can help you to learn how to create work of the same calibre. With practice, your skills will inevitably rise — but at any given time, it’s important to know the level you should aim for.
•    Find people who are cooler than you. Don’t hesitate to ask them for advice or to shadow them. Have them share their experiences, stand behind them and observe their work if they’ll let you. Few things are more useful than real work experience, both your own and that of others.
•    Take on a project. This could be a an illustrated workbook for colleagues or students, a guide book, a lecture for schoolchildren with compelling visuals, a course for students or a documentary on your topic.
•    If you work in a team, you can raise the bar even higher. Use each other’s strengths, share experiences, make plans, apply for grants and take on challenging science-communication projects together. This multiplies the fun and the results. More

Ceballos, G., García, A. & Ehrlich, P. R. The sixth extinction crisis: Loss of animal populations and species. J. Cosmol. 8, 31 (2010).
Google Scholar 
Johnson, C. N. et al. Biodiversity losses and conservation responses in the Anthropocene. Science 356, 270–275 (2017).CAS 
PubMed 

Google Scholar 
Scott, J. M., Goble, D. D., Haines, A. M., Wiens, J. A. & Neel, M. C. Conservation-reliant species and the future of conservation. Conserv. Lett. 3, 91–97 (2010).
Google Scholar 
Johnson, M. A., Kirby, R., Wang, S. & Losos, J. What drives variation in habitat use by Anolis lizards: Habitat availability or selectivity?. Can. J. Zool. 84, 877–886 (2006).
Google Scholar 
Gaston, K. J., Blackburn, T. M. & Lawton, J. H. Interspecific abundance-range size relationships: an appraisal of mechanisms. J. Anim. Ecol. 66, 579–601 (1997).
Google Scholar 
Devictor, V. et al. Defining and measuring ecological specialization. J. Appl. Ecol. 47, 15–25 (2010).
Google Scholar 
Razgour, O., Hanmer, J. & Jones, G. Using multi-scale modelling to predict habitat suitability for species of conservation concern: The grey long-eared bat as a case study. Biol. Cons. 144, 2922–2930 (2011).
Google Scholar 
Jetz, W., Sekercioglu, C. H. & Watson, J. E. Ecological correlates and conservation implications of overestimating species geographic ranges. Conserv. Biol. 22, 110–119 (2008).PubMed 

Google Scholar 
Seddon, P. J. From reintroduction to assisted colonization: Moving along the conservation translocation spectrum. Restor. Ecol. 18, 796–802 (2010).
Google Scholar 
Tomlinson, S., Lewandrowski, W., Elliott, C. P., Miller, B. P. & Turner, S. R. High-resolution distribution modeling of a threatened short-range endemic plant informed by edaphic factors. Ecol. Evol. 10, 763–773 (2019).PubMed 
PubMed Central 

Google Scholar 
Tomlinson, S., Webber, B. L., Bradshaw, S. D., Dixon, K. W. & Renton, M. Incorporating biophysical ecology into high-resolution restoration targets: insect pollinator habitat suitability models. Restor. Ecol. 26, 338–347 (2018).
Google Scholar 
Glen, A. S., Sutherland, D. R. & Cruz, J. An improved method of microhabitat assessment relevant to predation risk. Ecol. Res. 25, 311–314 (2010).
Google Scholar 
Limberger, D., Trillmich, F., Biebach, H. & Stevenson, R. D. Temperature regulation and microhabitat choice by free-ranging Galapagos fur seal pups (Arctocephalus galapagoensis). Oecologia 69, 53–59 (1986).PubMed 

Google Scholar 
Parmenter, R. R., Parmenter, C. A. & Cheney, C. D. Factors influencing microhabitat partitioning in arid-land darkling beetles (Tenebrionidae): temperature and water conservation. J. Arid Environ. 17, 57–67 (1989).
Google Scholar 
Kleckova, I., Konvicka, M. & Klecka, J. Thermoregulation and microhabitat use in mountain butterflies of the genus Erebia: importance of fine-scale habitat heterogeneity. J. Therm. Biol 41, 50–58 (2014).PubMed 

Google Scholar 
Napierała, A. & Błoszyk, J. Unstable microhabitats (merocenoses) as specific habitats of Uropodina mites (Acari: Mesostigmata). Exp. Appl. Acarol. 60, 163–180 (2013).PubMed 
PubMed Central 

Google Scholar 
Marshall, K. L., Philpot, K. E. & Stevens, M. Microhabitat choice in island lizards enhances camouflage against avian predators. Sci. Rep. 6, 1–10 (2016).
Google Scholar 
Lovell, P. G., Ruxton, G. D., Langridge, K. V. & Spencer, K. A. Egg-laying substrate selection for optimal camouflage by quail. Curr. Biol. 23, 260–264 (2013).CAS 
PubMed 

Google Scholar 
Wrege, P. H., Rowland, E. D., Keen, S. & Shiu, Y. Acoustic monitoring for conservation in tropical forests: Examples from forest elephants. Methods Ecol. Evol. 8, 1292–1301 (2017).
Google Scholar 
Measey, G. J., Stevenson, B. C., Scott, T., Altwegg, R. & Borchers, D. L. Counting chirps: Acoustic monitoring of cryptic frogs. J. Appl. Ecol. 54, 894–902 (2017).
Google Scholar 
Lambert, K. T. & McDonald, P. G. A low-cost, yet simple and highly repeatable system for acoustically surveying cryptic species. Austral Ecol. 39, 779–785 (2014).
Google Scholar 
Picciulin, M., Kéver, L., Parmentier, E. & Bolgan, M. Listening to the unseen: Passive Acoustic Monitoring reveals the presence of a cryptic fish species. Aquat. Conserv. Mar. Freshwat. Ecosyst. 29, 202–210 (2019).
Google Scholar 
Linkie, M. et al. Cryptic mammals caught on camera: assessing the utility of range wide camera trap data for conserving the endangered Asian tapir. Biol. Cons. 162, 107–115 (2013).
Google Scholar 
Balme, G. A., Hunter, L. T. & Slotow, R. Evaluating methods for counting cryptic carnivores. J. Wildl. Manag. 73, 433–441 (2009).
Google Scholar 
Carbone, C. et al. The use of photographic rates to estimate densities of tigers and other cryptic mammals in Animal Conservation forum. 75–79 (2001) (Cambridge University Press).Russell, J. C., Hasler, N., Klette, R. & Rosenhahn, B. Automatic track recognition of footprints for identifying cryptic species. Ecology 90, 2007–2013 (2009).PubMed 

Google Scholar 
Jarvie, S. & Monks, J. Step on it: can footprints from tracking tunnels be used to identify lizard species?. N. Z. J. Zool. 41, 210–217 (2014).
Google Scholar 
Watts, C., Thornburrow, D., Rohan, M. & Stringer, I. Effective monitoring of arboreal giant weta (Deinacrida heteracantha and D. mahoenui; Orthoptera: Anostostomatidae) using footprint tracking tunnels. J. Orthop. Res. 22, 93–100 (2013).
Google Scholar 
Williams, E. M. Developing monitoring methods for cryptic species: a case study of the Australasian bittern, Botaurus poiciloptilus: a thesis presented in partial fulfilment of the requirements for the degree of Doctor of Philosophy in Ecology at Massey University, Manawatū, New Zealand, Massey University (2016).Hacking, J., Abom, R. & Schwarzkopf, L. Why do lizards avoid weeds?. Biol. Invasions 16, 935–947 (2014).
Google Scholar 
Valentine, L. E. Habitat avoidance of an introduced weed by native lizards. Austral. Ecol. 31, 732–735 (2006).
Google Scholar 
Hawkins, J. P., Roberts, C. M. & Clark, V. The threatened status of restricted-range coral reef fish species in Animal Conservation forum. 81–88 (2000) (Cambridge University Press).Mason, L. D., Bateman, P. W. & Wardell-Johnson, G. W. The pitfalls of short-range endemism: High vulnerability to ecological and landscape traps. PeerJ 6, e4715 (2018).PubMed 
PubMed Central 

Google Scholar 
Dassot, M., Constant, T. & Fournier, M. The use of terrestrial LiDAR technology in forest science: Application fields, benefits and challenges. Ann. For. Sci. 68, 959–974 (2011).
Google Scholar 
Weber, H. LiDAR Sensor Functionality and Variants (2018).Michel, P., Jenkins, J., Mason, N., Dickinson, K. & Jamieson, I. Assessing the ecological application of lasergrammetric techniques to measure fine-scale vegetation structure. Eco. Inform. 3, 309–320 (2008).
Google Scholar 
Lim, K., Treitz, P., Wulder, M., St-Onge, B. & Flood, M. LiDAR remote sensing of forest structure. Prog. Phys. Geogr. 27, 88–106 (2003).
Google Scholar 
Anderson, L. & Burgin, S. Patterns of bird predation on reptiles in small woodland remnant edges in peri-urban north-western Sydney, Australia. Landsc. Ecol. 23, 1039–1047 (2008).
Google Scholar 
Hannam, M. & Moskal, L. M. Terrestrial laser scanning reveals seagrass microhabitat structure on a tideflat. Remote Sensing 7, 3037–3055 (2015).
Google Scholar 
Zavalas, R., Ierodiaconou, D., Ryan, D., Rattray, A. & Monk, J. Habitat classification of temperate marine macroalgal communities using bathymetric LiDAR. Remote Sens. 6, 2154–2175 (2014).
Google Scholar 
Mandlburger, G., Hauer, C., Wieser, M. & Pfeifer, N. Topo-bathymetric LiDAR for monitoring river morphodynamics and instream habitats—A case study at the Pielach River. Remote Sens. 7, 6160–6195 (2015).
Google Scholar 
Laize, C. et al. Use of LIDAR to characterise river morphology (2014).Cooper, C. & Withers, P. Physiological significance of the microclimate in night refuges of the numbat Myrmecobius fasciatus. Austral. Mammal. 27, 169–174 (2005).
Google Scholar 
Orell, P. & Morris, K. Chuditch recovery plan. Western Austral. Wildl. Manag. Program 13, 1 (1994).
Google Scholar 
Pearson, D. Western Spiny-Tailed Skink (Egernia stokesii) Recovery Plan (Department of Environment and Conservation, 2012).
Google Scholar 
McPeek, M. A., Cook, B. & McComb, W. Habitat selection by small mammals. Trans. Kentucky Acad. Sci. 44, 68–73 (1983).
Google Scholar 
Armstrong, K. The distribution and roost habitat of the orange leaf-nosed bat, Rhinonicteris aurantius, in the Pilbara region of Western Australia. Wildl. Res. 28, 95–104 (2001).
Google Scholar 
Mancina, C. et al. Endemics under threat: an assessment of the conservation status of Cuban bats. Hystrix Ital. J. Mammal. 18, 3–15 (2007).
Google Scholar 
Webb, M. H., Holdsworth, M. C. & Webb, J. Nesting requirements of the endangered Swift Parrot (Lathamus discolor). Emu-Austral. Ornithol. 112, 181–188 (2012).
Google Scholar 
Watson, S. J., Watson, D. M., Luck, G. W. & Spooner, P. G. Effects of landscape composition and connectivity on the distribution of an endangered parrot in agricultural landscapes. Landsc. Ecol. 29, 1249–1259 (2014).
Google Scholar 
Duffield, G. & Bull, M. Stable social aggregations in an Australian lizard, Egernia stokesii. Naturwissenschaften 89, 424–427 (2002).CAS 
PubMed 

Google Scholar 
Duffield, G. A. & Bull, M. Characteristics of the litter of the gidgee skink, Egernia stokesii. Wildl. Res. 23, 337–341 (1996).
Google Scholar 
Ecoscape. Blue Hills – Mungada East Terrestrial Fauna Assessment. (Sinosteel Midwest Corporation, 2016).Silver Lake Resources. Department of Water and Environmental Regulation Prescribe Premise Licence Application. (Egan Street Resources Limited, 2021).Maptek. I-Site 8800 Scanning System Solutions for Mining (2010).SoilWater Group. 3D LiDAR Scanning (2018).United States Department of Transportation. Ground-Based LiDAR Rock Slope Mapping and Assessment (2008).R Core Team. R: a language and environment for statistical computing, https://www.R-project.org/ (2017).Bartoń, K. Package ‘MuMIn’, https://cran.r-project.org/web/packages/MuMIn/MuMIn.pdf (2020).Converse, S. J., White, G. C. & Block, W. M. Small mammal responses to thinning and wildfire in ponderosa pine-dominated forests of the southwestern United States. J. Wildl. Manag. 70, 1711–1722 (2006).
Google Scholar 
Vieira, I. C. G. et al. Classifying successional forests using Landsat spectral properties and ecological characteristics in eastern Amazonia. Remote Sens. Environ. 87, 470–481 (2003).
Google Scholar 
Whitford, K. & Williams, M. Hollows in jarrah (Eucalyptus marginata) and marri (Corymbia calophylla) trees: II. Selecting trees to retain for hollow dependent fauna. For. Ecol. Manag. 160, 215–232 (2002).
Google Scholar 
Salmona, J., Dixon, K. M. & Banks, S. C. The effects of fire history on hollow-bearing tree abundance in montane and subalpine eucalypt forests in southeastern Australia. For. Ecol. Manag. 428, 93–103 (2018).
Google Scholar 
Lindenmayer, D., Cunningham, R., Donnelly, C., Tanton, M. & Nix, H. The abundance and development of cavities in Eucalyptus trees: a case study in the montane forests of Victoria, southeastern Australia. For. Ecol. Manage. 60, 77–104 (1993).
Google Scholar 
Craig, M. D. et al. How many mature microhabitats does a slow-recolonising reptile require? Implications for restoration of bauxite minesites in south-western Australia. Aust. J. Zool. 59, 9–17 (2011).
Google Scholar 
Schwarzkopf, L., Barnes, M. & Goodman, B. Belly up: Reduced crevice accessibility as a cost of reproduction caused by increased girth in a rock-using lizard. Austral Ecol. 35, 82–86 (2010).
Google Scholar 
Cooper, W. E. Jr. & Whiting, M. J. Islands in a sea of sand: Use of Acacia trees by tree skinks in the Kalahari Desert. J. Arid Environ. 44, 373–381 (2000).
Google Scholar 
Webb, J. K. & Shine, R. Out on a limb: conservation implications of tree-hollow use by a threatened snake species (Hoplocephalus bungaroides: Serpentes, Elapidae). Biol. Cons. 81, 21–33 (1997).
Google Scholar 
Fitzgerald, M., Shine, R. & Lemckert, F. Radiotelemetric study of habitat use by the arboreal snake Hoplocephalus stephensii (Elapidae) in eastern Australia. Copeia 2002, 321–332 (2002).
Google Scholar 
Grimm-Seyfarth, A., Mihoub, J. B. & Henle, K. Too hot to die? The effects of vegetation shading on past, present, and future activity budgets of two diurnal skinks from arid Australia. Ecol. Evol. 7, 6803–6813 (2017).PubMed 
PubMed Central 

Google Scholar 
Attum, O., Eason, P., Cobbs, G. & El Din, S. M. B. Response of a desert lizard community to habitat degradation: Do ideas about habitat specialists/generalists hold?. Biol. Cons. 133, 52–62 (2006).
Google Scholar 
Melville, J. & Schulte Ii, J. A. Correlates of active body temperatures and microhabitat occupation in nine species of central Australian agamid lizards. Austral. Ecol. 26, 660–669. https://doi.org/10.1046/j.1442-9993.2001.01152.x (2001).Article 

Google Scholar 
Munguia-Vega, A., Rodriguez-Estrella, R., Shaw, W. W. & Culver, M. Localized extinction of an arboreal desert lizard caused by habitat fragmentation. Biol. Cons. 157, 11–20 (2013).
Google Scholar 
Pietrek, A., Walker, R. & Novaro, A. Susceptibility of lizards to predation under two levels of vegetative cover. J. Arid Environ. 73, 574–577 (2009).
Google Scholar 
Moreno, S., Delibes, M. & Villafuerte, R. Cover is safe during the day but dangerous at night: The use of vegetation by European wild rabbits. Can. J. Zool. 74, 1656–1660 (1996).
Google Scholar 
Tchabovsky, A. V., Krasnov, B., Khokhlova, I. S. & Shenbrot, G. I. The effect of vegetation cover on vigilance and foraging tactics in the fat sand rat Psammomys obesus. J. Ethol. 19, 105–113 (2001).
Google Scholar 
Pizzuto, T. A., Finlayson, G. R., Crowther, M. S. & Dickman, C. R. Microhabitat use by the brush-tailed bettong (Bettongia penicillata) and burrowing bettong (B. lesueur) in semiarid New South Wales: Implications for reintroduction programs. Wildl. Res. 34, 271–279 (2007).
Google Scholar 
Hawlena, D., Saltz, D., Abramsky, Z. & Bouskila, A. Ecological trap for desert lizards caused by anthropogenic changes in habitat structure that favor predator activity. Conserv. Biol. 24, 803–809 (2010).PubMed 

Google Scholar 
Oversby, W., Ferguson, S., Davis, R. A. & Bateman, P. Bad news for bobtails: Understanding predatory behaviour of a resource-subsidised corvid towards an island endemic reptile. Wildl. Res. 45, 595–601 (2018).
Google Scholar 
Pianka, E. R. Rarity in A ustralian desert lizards. Austral Ecol. 39, 214–224 (2014).
Google Scholar 
Germano, J. M. & Bishop, P. J. Suitability of amphibians and reptiles for translocation. Conserv. Biol. 23, 7–15 (2009).PubMed 

Google Scholar 
Tsiouvaras, C., Havlik, N. & Bartolome, J. Effects of goats on understory vegetation and fire hazard reduction in a coastal forest in California. For. Sci. 35, 1125–1131 (1989).
Google Scholar 
Tasker, E. M. & Bradstock, R. A. Influence of cattle grazing practices on forest understorey structure in north-eastern New South Wales. Austral. Ecol. 31, 490–502 (2006).
Google Scholar 
Payne, A., Van Vreeswyk, A., Leighton, K., Pringle, H. & Hennig, P. An inventory and condition survey of the Sandstone-Yalgoo-Paynes Find area, Western Australia (1998).Shoo, L. P., Freebody, K., Kanowski, J. & Catterall, C. P. Slow recovery of tropical old-field rainforest regrowth and the value and limitations of active restoration. Conserv. Biol. 30, 121–132 (2016).PubMed 

Google Scholar 
Lamb, D. in Regreening the Bare Hills 325–358 (Springer, 2011).Bowler, D. E. & Benton, T. G. Causes and consequences of animal dispersal strategies: Relating individual behaviour to spatial dynamics. Biol. Rev. 80, 205–225 (2005).PubMed 

Google Scholar 
Stow, A. J., Sunnucks, P., Briscoe, D. & Gardner, M. The impact of habitat fragmentation on dispersal of Cunningham’s skink (Egernia cunninghami): Evidence from allelic and genotypic analyses of microsatellites. Mol. Ecol. 10, 867–878 (2001).CAS 
PubMed 

Google Scholar 
Stow, A. & Sunnucks, P. High mate and site fidelity in Cunningham’s skinks (Egernia cunninghami) in natural and fragmented habitat. Mol. Ecol. 13, 419–430 (2004).CAS 
PubMed 

Google Scholar  More

Direct habitat lossAccording to the global projections of urban expansion under five SSPs17 (Supplementary Note 3 and Supplementary Fig. 1), 36–74 million hectares (Mha) of land areas will be urbanized by 2100, representing a 54–111% increase compared with the baseline year of 2015. Among these, 11–33 Mha natural habitats (Supplementary Table 1) will become urban areas by 2100. Across SSP scenarios, the patterns of change in losses of total habitat, forest, shrubland, and grassland are consistent with the global projections of urban expansion (Fig. 1). In terms of urban encroachment on wetlands, wetland will undergo the largest loss under scenario SSP4 than under other scenarios. However, if the sustainable pathway of scenario SSP1 is properly implemented, this will enable us to conserve the global wetland. The greatest loss of other habitat will occur under scenario SSP3, but the minimal loss of other habitat will occur under scenario SSP1. Under the five different SSP scenarios, the United States, Nigeria, Australia, Germany, and the UK are consistently predicted to have greater habitat loss due to urban expansion (Supplementary Table 2).Fig. 1: Future direct habitat loss due to urban expansion under SSP scenarios.a The habitat loss by 2100 for each habitat type. Bars indicate the mean habitat loss area (five scenarios) for each habitat type. Error bars represent mean values ± 1 SEM for the loss of each habitat type under five scenarios, n = 5 scenarios. Points represent data in five scenarios. b The losses in total area, forest, shrubland, grassland, wetland, and other land.Full size imageThere are obvious disparities in the hot spots and cold spots of habitat loss under the five SSP scenarios (Fig. 2 and Supplementary Figs. 2–6). Potential hot spots of habitat loss are concentrated in regions such as the northeastern, southern, and western coasts of the United States, the Gulf of Guinea coastal areas, Sub-Saharan Africa, and the Persian Gulf coastal areas. Under scenario SSP5, parts of central and western Europe will also become hot spots. However, under other scenarios, the cold spots will be particularly concentrated in eastern and southern Europe. East Asia and South Asia, which are represented by China, India, and Japan, are dominated by cold spots (Supplementary Figs. 2–6), because these regions may experience a decline in urban land demand from 2050 to 2100 (for examples in China, see Supplementary Figs. 7–11), although they are currently the most populous regions in the world.Fig. 2: Future hot spots and cold spots of habitat loss due to urban expansion under SSP scenarios by 2100.Figures for the United States (a), Europe (b), Africa (c), and China (d) are presented separately. The Gi_Bin identifies statistically significant hot spots and cold spots. Statistical significance was based on the p-value and z-score (two-sided), and no adjustments were made for multiple comparisons.Full size imageOur scenario projections show that the largest natural habitat loss is expected to occur in the temperate broadleaf and mixed forests biome (except for scenario SSP3). In addition, many biomes will experience proportionate loss of natural habitat. These biomes include the tropical and subtropical coniferous forests biome, the temperate coniferous forests biome, the flooded grasslands and savannas biome, the Mediterranean forests, woodlands, and scrub biome, and the mangroves biome (Supplementary Table 3). Although the rate of future habitat loss is small at the global scale, it can be large in some areas. For example, the habitat in the temperate broadleaf and mixed forests may decrease by 1.4% under scenario SSP5. At the ecoregion scale, about 9% of 867 terrestrial ecoregions will lose more than 1% of habitat due to urban expansion (Supplementary Fig. 12). In the future, four ecoregions—the Atlantic coastal pine barrens, the coastal forests of the northeastern United States, and the Puerto Rican moist and dry forests—will experience more than 20% of habitat loss.Urban expansion threatens biodiversity prioritization schemesTo reflect the potential impact of urban expansion on protected areas (Supplementary Note 4), the analyses presented here were based on the assumption that urban expansion within protected areas is not strictly restricted and can even occur in the currently gazetted protected areas (Supplementary Note 5, Supplementary Figs. 13 and 14). In 2015, urban areas with a total area of 30,594 km2 were distributed in 28,152 protected areas, accounting for 12.6% of global protected areas (Supplementary Figs. 15 and 16). Moreover, 38% of the urban land-use changes within protected areas were due to the conversion of natural habitats into urban land between 1992 and 2015. If urban expansion continues without strict restrictions, 13.2–19.8% of the protected areas will be affected by urban land by 2100, and urban land will occur in 29,563–44,400 protected areas with a total urban land area of up to 46,705–89,901 km2 across the five SSP scenarios (the lowest and highest proportions of urban land in each protected area by 2100 under SSP3 and SSP5 scenarios are presented in Supplementary Figs. 17 and 18).We also found that 0.90% of all terrestrial biodiversity hotspots (Supplementary Note 6), which are the world’s most biologically rich yet threatened terrestrial regions24, were urbanized in 2015. And this proportion (0.90%) is higher than that located in the rest of the Earth’s surface (0.51%) in 2015. By 2100, the new urban expansion will additionally occupy 1.5–1.8% of hotspot areas under the five SSP scenarios (Supplementary Table 4). Five biodiversity hotspots are projected to suffer the largest proportion of urban land conversion: the California Floristic Province (6–11%), Japan (6–8%), the North American Coastal Plain (4–8%), the Guinean Forests of West Africa (4–8%), and the Forests of East Australia (2–6%). In contrast, the East Melanesian Islands and the New Caledonia are almost unaffected by urban expansion. Biodiversity hotspots (e.g., the Guinean Forests of West Africa, the Coastal Forests of Eastern Africa, Eastern Afromontane, and the Polynesia-Micronesia) with few human disturbances in 2015 are projected to experience the highest percentage of future urban growth. Compared with the urban areas in 2015, by 2100, the urban areas in these four biodiversity hotspots will experience a disproportionate increase of 281–708, 294–535, 169–305, and 33–337%, respectively.The World Wildlife Fund (WWF) selected the ecoregions that are most crucial to the conservation of global biodiversity as Global 20025 (Supplementary Note 7). However, about 93% of the Global 200 ecoregions will be affected by future urban expansion. Although the proportion of urban land in each ecoregion will be less than 1% in 2100, the urban area located in these ecoregions will experience an increase of 74–160% from 2015 to 2100 across the five SSP scenarios (Supplementary Table 4). Four ecologically vulnerable ecoregions that have the highest urban growth rates are the Sudd-Sahelian Flooded Grasslands and Savannas, the East African Acacia Savannas, the Hawaii Moist Forest, and the Congolian Coastal Forests. By 2100, the urban areas in these four ecoregions will increase by 877–9955, 527–646, 18–902, and 500–1037%, respectively.The five SSP scenarios showed that the urban area is expected to increase by only 73–213 km2 in the Last of the Wild areas26 (see Supplementary Note 8 for descriptions about the Last of the Wild areas) by 2100 (Supplementary Table 4).Impacts of urban expansion on habitat fragmentationThe increasing exposures of natural habitat to urbanized land use may cause long-term changes in the function and structure of the natural habitat that is adjacent to urban areas13. To examine this proximity effect, we investigated the impact of future urban expansion on the nearest distance between urban areas and natural habitat (i.e., the distance from patch edges of urban areas to patch edges of the nearest natural habitats) under different SSP scenarios. Although the global urban area is expected to increase by 36–74 Mha by 2100, the impacts of future urban expansion on adjacent natural habitat are disproportionately large. Future urban expansion will make urban areas much closer to patch edges of 34–40 Mha natural habitat, which will inevitably threaten the natural habitat and increase the risk of biodiversity decline. The effects of urban expansion on adjacent patch edges of natural habitats are remarkably different across different scenarios. Specifically, the area of affected adjacent natural habitat is expected to be 38.45, 34.24, 40.31, 37.84, and 39.42 Mha under SSP1 to SSP5 scenarios by 2100, with the smallest effect under scenario SSP2, and the largest effect under scenario SSP3. Moreover, the scale of urban expansion does not correspond directly with the size of the impact. Several countries, including Mauritania, Algeria, Saudi Arabia, Western Sahara, and the United States, will have a large change in the distance from future urban areas to natural habitats due to urban expansion (Supplementary Table 5). Such effects also varied across different natural habitat types. The distance from the patch edges of urban areas to patch edges of (a) wetland, other land, and forest, (b) grassland, and (c) shrubland will generally be shortened by ~2000, ~1500 and ~900 m, respectively.In addition to the effect on the distance to the habitat edge, urban-caused habitat fragmentation is also reflected in reducing mean patch size (MPS)13, increasing mean edge index (edge density (ED), i.e., edge length on a per-unit area)27, and enlarging isolation (mean Euclidean nearest neighbor distance, ENN_MN)28 (Fig. 3). Taking the global ecoregions as the analysis unit, we found that within a 5 km buffer of urban areas, the median of MPS of natural habitats tends to show an overall decline trend, and the segmentation and subdivision of habitats become more obvious as future urban land expands. The median of MPS is the largest under scenario SSP1, followed by SSP4, SPP2, and SSP3 with some fluctuations in between, and the smallest MPS is found with the most fragmented landscape under scenario SSP5. A smaller patch size indicates that the inner parts of the habitat are subject to higher risk of being influenced by external disturbance. Future urban expansion also tends to cause an increase in the ED of natural habitat, which is often linked with smaller patches or more irregular shapes, and therefore poses a threat to biodiversity that influences many ecological processes (e.g., the spread of dispersal and predation)13,27,28. Scenario SSP1 shows the best performance in maintaining a low habitat ED and a high level of biodiversity conservation. However, under scenario SSP5, ED will experience a rapid increase in the second half of the 21st century. Meanwhile, the ENN_MN will increase substantially in the future, suggesting that areas with the same habitat type will become increasingly isolated, irregular, dispersed, or unevenly distributed due to the barrier of urban land. This will affect the speed of dispersal and patch recolonization. Scenario SSP1 is also most conducive to maintaining the proximity of natural habitats with the same habitat type. Other scenarios show relatively similar performance.Fig. 3: Future urban expansion effects on habitat fragmentation under SSP scenarios.a Mean patch size (MPS), b edge density (ED), c mean Euclidean nearest-neighbor distance (ENN_MN).Full size imageImpacts of urban expansion on terrestrial biodiversityWe focus on biodiversity in three common vertebrate taxa (i.e., amphibians, mammals, and birds) in our analyses. Future land system conversion to urban land will cause an average of 34% loss in the overall relative species richness. Land conversion from dense forest, mosaic grassland and open forest, mosaic grassland, and bare and natural grassland to urban land will cause the highest overall relative biodiversity loss (48%, 95% confidence interval (CI): 34–59% on a 1 km grid). These land systems with a high risk of biodiversity loss are concentrated in the United States, Europe, and Sub-Saharan Africa (Supplementary Fig. 19). Overall, the negative effect of future urban expansion on the total abundance of species will be more pronounced than that on species richness. Urban land changes will result in an average of 52% overall loss in relative total abundance of species. In particular, the losses of dense forest, natural grassland, and mosaic grassland, due to conversion to urban land, will lead to a high risk of species loss (62%, 95% CI: 38–76%).In terms of the number of species (i.e., all amphibians, mammals, and birds), future urban expansion will cause an average loss of 7–9 species and a loss of up to ~197 species per 10 km grid cell by 2100 across the five SSP scenarios (Fig. 4 and Supplementary Fig. 20). Species loss is most likely to be concentrated in Sub-Saharan Africa (particularly the Gulf of Guinea coast), the United States, and Europe. In addition, southeastern Brazil, India, and the eastern coast of Australia are also relatively high-risk areas. However, the specific effects of urban expansion vary substantially across different SSP scenarios. For instance, under scenario SSP5, urban expansion will pose a fatal threat to the global species richness in areas with urban development potential (species richness loss will occur in ~740 Mha land areas), whereas under the divided pathway (SSP4) and regional rivalry pathway (SSP3) scenarios, urban expansion will threaten the richest biodiversity hotspots, such as Sub-Saharan Africa and Latin America (Supplementary Fig. 20).Fig. 4: Potential biodiversity loss due to future urban expansion under SSP scenarios.The biodiversity loss in terms of the number of terrestrial vertebrate species (amphibians, mammals, and birds) lost per 10 km grid cell in the North America (a), Europe (b), the Gulf of Guinea coast (c), and East Asia (d).Full size imageWe also found a loss of up to 12 species of threatened amphibians, mammals, and birds (including vulnerable, endangered, or critically endangered categories defined in the IUCN Red List), and a loss of up to 40 species of small-ranged amphibians, mammals, and birds (small-ranged species are species with a geographic range size smaller than the median range size for that taxon)29 due to future urban expansion by 2100. There are a few scattered areas that will be hotspots for the loss of threatened species, such as West Africa, East Africa, northern India, and the eastern coast of Australia (Supplementary Fig. 21). The loss of small-ranged species will concentrate in fewer areas (Supplementary Fig. 22). We have identified 30 conservation priority ecoregions with high risks of habitat loss and small-ranged species loss due to future urban expansion (Supplementary Table 6). These conservation priority ecoregions are all found in Latin America and Sub-Saharan Africa (Supplementary Fig. 23). However, some hotspots outside of these conservation priority regions, such as tropical Southeast Asia, the west coast of the United States, and northern New Zealand, will also be affected (Supplementary Fig. 23).The top 5% 10 km grid cells with the highest loss in species richness (28–38 species potentially being lost) scatter across adjacent urban areas. However, only 6.4–8.6% of these regions are covered by the current global network of protected areas. These areas are often overlooked, and thus receive relatively low conservation spending. Ecoregions in Sub-Saharan African, Central and South America, Southeast Asia, and Australia will be responsible for the top 43% of average species loss across the SSP scenarios (Fig. 5). Kenya, Swaziland, Brunei, Zambia, Republic of Congo, and Zimbabwe will face the largest potential species richness loss (approximately > 29 species lost per 10 km grid cell) under all five SSP scenarios (Supplementary Fig. 24 and Supplementary Table 7).Fig. 5: Average potential biodiversity loss per 10 km grid cell in ecoregions due to future urban expansion under SSP scenarios.The mean potential biodiversity loss represents the average number of terrestrial vertebrate species (amphibians, mammals, and birds) lost per 10 km grid cell.Full size image More

Quicke, D. L. Parasitic Wasps (Chapman & Hall Ltd., 1997).
Google Scholar 
Godfray, H. C. J. Parasitoids: Behavioral and Evolutionary Ecology (Princeton University Press, 1994).
Google Scholar 
Mayhew, P. J. & van Alphen, J. J. M. Gregarious development in alysiine parasitoids evolved through a reduction in larval aggression. Anim. Behav. 58 , 131–141 (1999).Mayhew, P. J. & Hardy, I. C. W. Nonsiblicidal behavior and the evolution of clutch size in bethylid wasps. Am. Nat. 151, 409–424 (1998).ADS 
CAS 
PubMed 

Google Scholar 
Schmidt, J. M. & Smith, J. J. B. Correlations between body angles and substrate curvature in the parasitoid wasp Trichogramma minutum: A possible mechanism of host radius measurement. J. Exp. Biol. 125, 271–285 (1986).
Google Scholar 
Boivin, G. & Baaren, J. The role of larval aggression and mobility in the transition between solitary and gregarious development in parasitoid wasps. Ecol. Lett. 3, 469–474 (2000).
Google Scholar 
Rosenheim, J. A., Wilhoit, L. R. & Armer, C. A. Influence of intraguild predation among generalist insect predators on the suppression of an herbivore population. Oecologia 96, 439–449 (1993).ADS 
PubMed 

Google Scholar 
Mayhew, P. J. The evolution of gregariousness in parasitoid wasps. Proc. R. Soc. Lond. B Biol. 265, 383–389 (1998).
Google Scholar 
Harvey, P. H. & Partridge, L. Murderous mandibles and black holes in hymenopteran wasps. Nature 326, 128–129 (1987).ADS 

Google Scholar 
Pexton, J. J. & Mayhew, P. J. Competitive interactions between parasitoid larvae and the evolution of gregarious development. Oecologia 141, 179–190 (2004).ADS 
PubMed 

Google Scholar 
Pexton, J. J. & Mayhew, P. J. Immobility: The key to family harmony? Trends Ecol. Evol. 16, 7–9 (2001).CAS 
PubMed 

Google Scholar 
Godfray, H. C. J. The evolution of clutch size in parasitic wasps. Am. Nat. 129, 221–233 (1987).
Google Scholar 
Laing, J. E. & Corrigan, J. E. Intrinsic competition between the gregarious parasite, Cotesia glomeratus and the solitary parasite Cotesia rubecula (Hymenoptera: Braconidae) for their host Artogeia rapae (Lepidoptera: Pieridae). Entomophaga 32, 493–501 (1987).
Google Scholar 
Pexton, J. J. & Mayhew, P. J. Clutch size adjustment, information use and the evolution of gregarious development in parasitoid wasps. Behav. Ecol. Soc. 58, 99–110 (2005).
Google Scholar 
Reitz, S. R. & Adler, P. H. Fecundity and oviposition of Eucelatoria bryani, a gregarious parasitoid of Helicoverpa zea and Heliothis virescens. Entomol. Exp. Appl. 75, 175–181 (1995).
Google Scholar 
Wei, K., Tang, Y. L., Wang, X. Y., Cao, L. M. & Yang, Z. Q. The developmental strategies and related profitability of an idiobiont ectoparasitoid Sclerodermus pupariae vary with host size. Ecol. Entomol. 39, 101–108 (2014).
Google Scholar 
van Alphen, J. J. M. & Visser, M. E. Superparasitism as an adaptive strategy for insect parasitoids. Ann. Rev. Entomol. 35, 59–79 (1990).
Google Scholar 
Mayhew, P. J. & Glaizot, O. Integrating theory of clutch size and body size evolution for parasitoids. Oikos 92, 372–376 (2001).
Google Scholar 
Samková, A., Hadrava, J., Skuhrovec, J. & Janšta, P. Reproductive strategy as a major factor determining female body size and fertility of a gregarious parasitoid. J. Appl. Entomol. 143, 441–450 (2019).
Google Scholar 
Hardy, I. C. W., Griffiths, N. T. & Godfray, H. C. J. Clutch size in a parasitoid wasp: A manipulation experiment. J. Anim. Ecol. 61, 121–129 (1992).
Google Scholar 
Visser, M. E. The importance of being large: The relationship between size and fitness in females of the parasitoid Aphaereta minuta (Hymenoptera: Braconidae). J. Anim. Ecol. 63, 963–978 (1994).
Google Scholar 
Sagarra, L. A., Vincent, C. & Stewart, R. K. Body size as an indicator of parasitoid quality in male and female Anagyrus kamali (Hymenoptera: Encyrtidae). Bull. Entomol. Res. 91, 363–367 (2001).CAS 
PubMed 

Google Scholar 
Bezemer, T. M. & Mills, N. J. Clutch size decisions of a gregarious parasitoid under laboratory and field conditions. Anim. Behav. 66, 1119–1128 (2003).
Google Scholar 
Takagi, M. The reproductive strategy of the gregarious parasitoid, Pteromalus puparum (Hymenoptera: Pteromalidae). Oecologia 68, 1–6 (1985).ADS 
PubMed 

Google Scholar 
Jervis, M. A., Ferns, P. N. & Heimpel, G. E. Body size and the timing of egg production in parasitoid wasps: A comparative analysis. Funct. Ecol. 17, 375–383 (2003).
Google Scholar 
Waage, J. K. & Lane, J. A. The reproductive strategy of a parasitic wasp: II. Sex allocation and local mate competition in Trichogramma evanescens. J. Anim. Ecol. 53, 417–426 (1984).
Google Scholar 
Waage, J. K. & Ming, N. S. The reproductive strategy of a parasitic wasp: I. Optimal progeny and sex allocation in Trichogramma evanescens. J. Anim. Ecol. 53, 401–415 (1984).
Google Scholar 
Rabinovich, J. E., Jorda, M. T. & Bernstein, C. Local mate competition and precise sex ratios in Telenomus fariai (Hymenoptera: Scelionidae), a parasitoid of triatomine eggs. Behav. Ecol. Sociobiol. 48, 308–315 (2000).
Google Scholar 
Goubault, M., Mack, A. F. & Hardy, I. C. W. Encountering competitors reduces clutch size and increases offspring size in a parasitoid with female–female fighting. Proc. R. Soc. B Biol. 274, 2571–2577 (2007).
Google Scholar 
Duval, J. F., Brodeur, J., Doyon, J. & Boivin, G. Impact of superparasitism time intervals on progeny survival and fitness of an egg parasitoid. Ecol. Entomol. 43, 310–317 (2018).
Google Scholar 
Mesterton-Gibbons, M. & Hardy, I. C. W. The influence of contests on optimal clutch size: A game–theoretic model. Proc. R. Soc. Lond. B Biol. 271, 971–978 (2004).
Google Scholar 
Koppik, M., Thiel, A. & Hoffmeister, T. S. Adaptive decision making or differential mortality: What causes offspring emergence in a gregarious parasitoid? Entomol. Exp. Appl. 150, 208–216 (2014).
Google Scholar 
Heimpel, G. E. Host–parasitoid population dynamics. In Parasitoid population biology (eds Hochberg, M. E. & Ives, A. R.) 27–40 (Princeton, 2000).
Google Scholar 
Zaviezo, T. & Mills, M. Factors influencing the evolution of clutch size in a gregarious insect parasitoid. J. Anim. Ecol. 69, 1047–1057 (2000).
Google Scholar 
Kazmer, D. J. & Luck, R. F. Field tests of the size-fitness hypothesis in the egg parasitoid Trichogramma pretiosum. Ecology 76, 412–425 (1995).
Google Scholar 
Segoli, M. & Rosenheim, J. A. The effect of body size on oviposition success of a minute parasitoid in nature. Ecol. Entomol. 40, 483–485 (2015).
Google Scholar 
Gao, S. K., Wei, K., Tang, Z. L., Wang, X. Y. & Yang, Z. Q. Effect of parasitoid density on the timing of parasitism and development duration of progeny in Sclerodermus pupariae (Hymenoptera: Bethylidae). Biol. Control 97, 57–62 (2016).
Google Scholar 
Anderson, R. C. & Paschke, J. D. The biology and ecology of Anaphes flavipes (Hymenoptera: Mymaridae), an exotic egg parasite of the cereal leaf beetle. Ann. Entomol. Soc. Am. 61, 1–5 (1968).
Google Scholar 
Hoffman, G. D. & Rao, S. Oviposition site selection on oats: The effect of plant architecture, plant and leaf age, tissue toughness, and hardness on cereal leaf beetle, Oulema melanopus. Entomol. Exp. Appl. 141, 232–244 (2011).
Google Scholar 
Samková, A., Hadrava, J., Skuhrovec, J. & Janšta, P. Host population density and presence of predators as key factors influencing the number of gregarious parasitoid Anaphes flavipes offspring. Sci. Rep. UK 9, 1–7 (2019).ADS 

Google Scholar 
Hardy, I. C. W. Sex ratio and mating structure in the parasitoid Hymenoptera. Oikos 69, 3–20 (1994).
Google Scholar 
Godfray, H. C. J. Models for clutch size and sex ratio with sibling interaction. Theor. Popul. Biol. 30, 215–231 (1986).MATH 

Google Scholar 
Hardy, I. C. W. Non-binomial sex allocation and brood sex ratio variances in the parasitoid Hymenoptera. Oikos 65, 143–158 (1992).
Google Scholar 
Petersen, G. & Hardy, I. C. W. The importance of being larger: Parasitoid intruder–owner contests and their implications for clutch size. Anim. Behav. 51, 1363–1373 (1996).
Google Scholar 
Klomp, H. & Teerink, B. J. The significance of oviposition rates in the egg parasite, Trichogramma embryophagum Htg. Arch. Neerl. Zool. 17, 350–375 (1967).
Google Scholar 
May, R. M., Hassell, M. P., Anderson, R. M. & Tonkyn, D. W. Density dependence in host–parasitoid models. J. Anim. Ecol. 50, 855–865 (1981).MathSciNet 

Google Scholar 
Hoddle, M. S., Van Driesche, R. G., Elkinton, J. S. & Sanderson, J. P. Discovery and utilization of Bemisia argentifolii patches by Eretmocerus eremicus and Encarsia formosa (Beltsville strain) in greenhouses. Entomol. Exp. Appl. 87, 15–28 (1998).
Google Scholar 
Samková, A., Raška, J., Hadrava, J. & Skuhrovec, J. Scarcity of hosts for gregarious parasitoids indicates an increase of individual offspring fertility by reducing their own fertility. bioRxiv https://doi.org/10.1101/2021.03.05.434037 (2021).Article 

Google Scholar 
van Dijken, M. J. & Waage, J. K. Self and conspecific superparasitism by the egg parasitoid Trichogramma evanescens. Entomol. Exp. Appl. 43, 183–192 (1987).
Google Scholar 
van de Vijver, E. et al. Inter-and intrafield distribution of cereal leaf beetle species (Coleoptera: Chrysomelidae) in Belgian winter wheat. Environ. Entomol. 48, 276–283 (2019).PubMed 

Google Scholar 
Samková, A., Hadrava, J., Skuhrovec, J. & Janšta, P. Host specificity of the parasitic wasp Anaphes flavipes (Hymenoptera: Mymaridae) and a new defence in its hosts (Coleoptera: Chrysomelidae: Oulema spp.). Insects 11, 175 (2020).PubMed Central 

Google Scholar 
Bezděk, J. & Baselga, A. Revision of western Palaearctic species of the Oulema melanopus group, with description of two new species from Europe (Coleoptera: Chrysomelidae: Criocerinae). Acta Entomol. Mus. Nat. Pragae 55, 273–304 (2015).
Google Scholar 
Anderson, R. C. & Paschke, J. D. Additional observations on the biology of Anaphes flavipes (Hymenoptera: Mymaridae), with special reference to the effects of temperature and superparasitism on development. Ann. Entomol. Soc. Am. 62, 1316–1321 (1969).
Google Scholar 
R Core Team. A Language and Environment for Statistical Computing. R Foundation for Statistical Computing (R Core Team, 2020).
Google Scholar 
Bates, D., Mächler, M., Bolker, B. & Walker, S. Fitting linear mixed-effects models using lme4. J. Stat. Softw. 67, 1–48 (2015). https://CRAN.R-project.org/package=lme4. More

Zhao, S. et al. Regulation of plant responses to salt stress. Int. J. Mol. Sci. 22, 4609 (2021).CAS 
PubMed 
PubMed Central 

Google Scholar 
Acosta-Motos, J. R. et al. Plant responses to salt stress: Adaptive mechanisms. Agronomy-Basel 7, 18 (2017).
Google Scholar 
Munns, R. & Tester, M. Mechanisms of salinity tolerance. Annu. Rev. Plant Biol. 59, 651–681 (2008).CAS 
PubMed 

Google Scholar 
Wu, H. H. Plant salt tolerance and Na+ sensing and transport. Crop J. 6, 215–225 (2018).
Google Scholar 
Ali, M. et al. Silicon mediated improvement in the growth and ion homeostasis by decreasing Na+ uptake in maize (Zea mays L.) cultivars exposed to salinity stress. Plant Physiol. Biochem. 158, 208–218 (2021).CAS 
PubMed 

Google Scholar 
Javaid, T., Farooq, M. A., Akhtar, J., Saqib, Z. A. & Anwar-ul-Haq, M. Silicon nutrition improves growth of salt-stressed wheat by modulating flows and partitioning of Na+, Cl- and mineral ions. Plant Physiol. Biochem. 141, 291–299 (2019).CAS 
PubMed 

Google Scholar 
Zelm, E. V., Zhang, Y. X. & Testerink, C. Salt tolerance mechanisms of plants. Annu. Rev. Plant Biol. 71, 403–433 (2020).PubMed 

Google Scholar 
Kumar, P. et al. Potassium: A key modulator for cell homeostasis. J. Biotechnol. 324, 198–210 (2020).CAS 
PubMed 

Google Scholar 
Ahmad, P., Ahanger, M. A., Alam, P., Alyemeni, M. N. & Ashraf, M. Silicon (Si) supplementation alleviates NaCl toxicity in mung bean [Vigna radiata (L.) Wilczek] through the modifications of physio-biochemical attributes and key antioxidant enzymes. J. Plant Growth Regul. 38, 1–13 (2018).
Google Scholar 
Chiappero, J. et al. Antioxidant status of medicinal and aromatic plants under the influence of growth-promoting rhizobacteria and osmotic stress. Ind. Crops Prod. 167, 113541 (2021).CAS 

Google Scholar 
Conceicao, S. S. et al. Silicon modulates the activity of antioxidant enzymes and nitrogen compounds in sunflower plants under salt stress. Arch. Agron. Soil Sci. 65, 1237–1247 (2019).CAS 

Google Scholar 
Etesami, H. & Jeong, B. R. Silicon (Si): Review and future prospects on the action mechanisms in alleviating biotic and abiotic stresses in plants. Ecotoxicol. Environ. Saf. 147, 881–896 (2018).CAS 
PubMed 

Google Scholar 
Epstein, E. Silicon. Annu. Rev. Plant Physiol. Plant Mol. Biol. 50, 641–664 (1999).CAS 
PubMed 

Google Scholar 
Epstein, E. The anomaly of silicon in plant biology. Proc. Natl. Acad. Sci. U S A 91, 11–17 (1994).CAS 
PubMed 
PubMed Central 

Google Scholar 
Jadhao, K. R., Bansal, A. & Rout, G. R. Silicon amendment induces synergistic plant defense mechanism against pink stem borer (Sesamia inferens Walker.) in finger millet (Eleusine coracana Gaertn.). Sci. Rep. 10, 15 (2020).
Google Scholar 
Li, Z. C. et al. Silicon enhancement of estimated plant biomass carbon accumulation under abiotic and biotic stresses. A meta-analysis. Agron. Sustain. Dev. 38, 19 (2018).CAS 

Google Scholar 
Yan, G. C. et al. Silicon improves rice salinity resistance by alleviating ionic toxicity and osmotic constraint in an organ-specific pattern. Front. Plant Sci. 11, 12 (2020).
Google Scholar 
Farouk, S., Elhindi, K. M. & Alotaibi, M. A. Silicon supplementation mitigates salinity stress on Ocimum basilicum L. via improving water balance, ion homeostasis, and antioxidant defense system. Ecotoxicol. Environ. Saf. 206, 11 (2020).
Google Scholar 
Yin, J. L. et al. Silicon enhances the salt tolerance of cucumber through increasing polyamine accumulation and decreasing oxidative damage. Ecotoxicol. Environ. Saf. 169, 8–17 (2019).CAS 
PubMed 

Google Scholar 
Hurtado, A. C. et al. Different methods of silicon application attenuate salt stress in sorghum and sunflower by modifying the antioxidative defense mechanism. Ecotoxicol. Environ. Saf. 203, 11 (2020).
Google Scholar 
Gaur, S. et al. Fascinating impact of silicon and silicon transporters in plants: A review. Ecotoxicol. Environ. Saf. 202, 12 (2020).
Google Scholar 
Vandegeer, R. K. et al. Silicon deposition on guard cells increases stomatal sensitivity as mediated by K(+)efflux and consequently reduces stomatal conductance. Physiol. Plant 171, 358–370 (2021).CAS 
PubMed 

Google Scholar 
Lina, et al. Silicon-mediated changes in polyamines participate in silicon-induced salt tolerance in Sorghum bicolor L.. Plant Cell Environ. 39, 245–258 (2016).
Google Scholar 
Hassanvand, F., Nejad, A. R. & Fanourakis, D. Morphological and physiological components mediating the silicon-induced enhancement of geranium essential oil yield under saline conditions. Ind. Crops Prod. 134, 19–25 (2019).CAS 

Google Scholar 
Altuntas, O., Dasgan, H. Y. & Akhoundnejad, Y. Silicon-induced salinity tolerance improves photosynthesis, leaf water status, membrane stability, and growth in pepper (Capsicum annuum L.). HortScience 53, 1820–1826 (2018).CAS 

Google Scholar 
Coskun, D. et al. The controversies of silicon’s role in plant biology. New Phytol. 221, 67–85 (2019).PubMed 

Google Scholar 
Jiang, M. Y. et al. An “essential herbal medicine”-licorice: A review of phytochemicals and its effects in combination preparations. J. Ethnopharmacol. 249, 14 (2020).
Google Scholar 
Zhang, X. Y. et al. Inhibition effect of glycyrrhiza polysaccharide (GCP) on tumor growth through regulation of the gut microbiota composition. J. Pharmacol. Sci. 137, 324–332 (2018).CAS 
PubMed 

Google Scholar 
Baltina, L. A. et al. Glycyrrhetinic acid derivatives as Zika virus inhibitors: Synthesis and antiviral activity in vitro. Bioorg. Med. Chem. 41, 116204 (2021).CAS 
PubMed 

Google Scholar 
Zhao, Z. Y. et al. Glycyrrhizic ccid nanoparticles as antiviral and anti-inflammatory agents for COVID-19 treatment. ACS Appl. Mater. Interfaces 13, 20995–21006 (2021).CAS 
PubMed 
PubMed Central 

Google Scholar 
Lu, J. H., Lv, X., Wu, L. & Li, X. Y. Germination responses of three medicinal licorices to saline environments and their suitable ecological regions. Acta Pratacul. Sin. 22, 198–205 (2013).
Google Scholar 
Geng, G. Q. & Xie, X. R. Effect of drought and salt stress on the physiological and biochemical characteristics of Glycyrrhiza uralensis. Pratacult. Sci. 35, 113–120 (2018).
Google Scholar 
Cui, J. J., Zhang, X. H., Li, Y. T., Zhou, D. & Zhang, E. H. Effect of silicon addition on seedling morphological and physiological indicators of Glycyrrhiza uralensis under salt stress. Acta Pratacul. Sin. 24, 214–220 (2015).
Google Scholar 
Zhang, W. J. et al. Silicon alleviates salt and drought stress of Glycyrrhiza uralensis plants by improving photosynthesis and water status. Biol. Plant. 64, 302–313 (2020).CAS 

Google Scholar 
Zhang, W. J. et al. Silicon promotes growth and root yield of Glycyrrhiza uralensis under salt and drought stresses through enhancing osmotic adjustment and regulating antioxidant metabolism. Crop Prot. 107, 1–11 (2018).
Google Scholar 
Chen, D. Q. et al. Silicon moderated the K deficiency by improving the plant-water status in sorghum. Sci. Rep. 6, 14 (2016).PubMed 
PubMed Central 

Google Scholar 
Cui, J. J., Zhang, E. H., Zhang, X. H. & Wang, Q. Silicon alleviates salinity stress in licorice (Glycyrrhiza uralensis) by regulating carbon and nitrogen metabolism. Sci. Rep. 11, 12 (2021).
Google Scholar 
Lichtenthaler, H. K. & Wellburn, A. R. Determinations of total carotenoids and chlorophylls a and b of leaf extracts in different solvents. Analysis 11, 591–592 (1983).CAS 

Google Scholar 
Yan, K., Wu, C. W., Zhang, L. H. & Chen, X. B. Contrasting photosynthesis and photoinhibition in tetraploid and its autodiploid honeysuckle (Lonicera japonica Thunb.) under salt stress. Front. Plant Sci. 6, 9 (2015).
Google Scholar 
Li, H. S. Principles and Techniques of Plant Physiological and Biochemical Experiments (Higher Education Press, 2000).
Google Scholar 
Bradford, M. M. A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal. Biochem. 72, 248–254 (1976).CAS 
PubMed 

Google Scholar 
Lutts, S., Kinet, J. M. & Bouharmont, J. NaCl-induced senescence in leaves of rice (Oryza sativa L) cultivars differing in salinity resistance. Ann. Bot. 78, 389–398 (1996).CAS 

Google Scholar 
Havir, E. A. & Mchale, N. A. Biochemical and developmental characterization of multiple forms of catalase in tobacco leaves. Plant Physiol. 84, 450–455 (1987).CAS 
PubMed 
PubMed Central 

Google Scholar 
Rizwan, M. et al. Mechanisms of silicon-mediated alleviation of drought and salt stress in plants: A review. Environ. Sci. Pollut. Res. 22, 15416–15431 (2015).CAS 

Google Scholar 
Al-Huqail, A. A., Alqarawi, A. A., Hashem, A., Malik, J. A. & Abd Allah, E. F. Silicon supplementation modulates antioxidant system and osmolyte accumulation to balance salt stress in Acacia gerrardii Benth. Saudi J. Biol. Sci. 26, 1856–1864 (2019).CAS 
PubMed 

Google Scholar 
Hurtado, A. C. et al. Silicon application induces changes C:N: P stoichiometry and enhances stoichiometric homeostasis of sorghum and sunflower plants under salt stress. Saudi J. Biol. Sci. 27, 3711–3719 (2020).
Google Scholar 
Zhang, X. H., Zhang, W. J., Lang, D. Y., Cui, J. J. & Li, Y. T. Silicon improves salt tolerance of Glycyrrhiza uralensis Fisch by ameliorating osmotic and oxidative stresses and improving phytohormonal balance. Environ. Sci. Pollut. Res. 25, 25916–25932 (2018).CAS 

Google Scholar 
Liang, W. J., Ma, X. L., Wan, P. & Liu, L. Y. Plant salt-tolerance mechanism: A review. Biochem. Biophys. Res. Commun. 495, 286–291 (2018).CAS 
PubMed 

Google Scholar 
Tester, M. & Davenport, R. Na+ tolerance and Na+ transport in higher plants. Ann. Bot. 91, 503–527 (2003).CAS 
PubMed 
PubMed Central 

Google Scholar 
Khan, W. U. D. et al. Silicon nutrition mitigates salinity stress in maize by modulating ion accumulation, photosynthesis, and antioxidants. Photosynthetica 56, 1047–1057 (2018).CAS 

Google Scholar 
Zahoor, R. et al. Potassium fertilizer improves drought stress alleviation potential in cotton by enhancing photosynthesis and carbohydrate metabolism. Environ. Exp. Bot. 137, 73–83 (2017).CAS 

Google Scholar 
Hurtado, A. C. et al. Silicon alleviates sodium toxicity in sorghum and sunflower plants by enhancing ionic homeostasis in roots and shoots and increasing dry matter accumulation. SILICON 13, 475–486 (2021).CAS 

Google Scholar 
Yan, G. C. et al. Silicon alleviates salt stress-induced potassium deficiency by promoting potassium uptake and translocation in rice (Oryza sativa L.). J. Plant Physiol. 258, 7 (2021).
Google Scholar 
Dhiman, P. et al. Fascinating role of silicon to combat salinity stress in plants: An updated overview. Plant Physiol. Biochem. 162, 110–123 (2021).CAS 
PubMed 

Google Scholar 
Bosnic, P., Bosnic, D., Jasnic, J. & Nikolic, M. Silicon mediates sodium transport and partitioning in maize under moderate salt stress. Environ. Exp. Bot. 155, 681–687 (2018).CAS 

Google Scholar 
Alamri, S. et al. Silicon-induced postponement of leaf senescence is accompanied by modulation of antioxidative defense and ion homeostasis in mustard (Brassica juncea) seedlings exposed to salinity and drought stress. Plant Physiol. Biochem. 157, 47–59 (2020).CAS 
PubMed 

Google Scholar 
Ahmad, P. et al. Nitric oxide mitigates salt stress by regulating levels of osmolytes and antioxidant enzymes in chickpea. Front. Plant Sci. 7, 1–11 (2016).
Google Scholar 
Zhu, Y. X. et al. Silicon confers cucumber resistance to salinity stress through regulation of proline and cytokinins. Plant Physiol. Biochem. 156, 209–220 (2020).CAS 
PubMed 

Google Scholar  More

Graded lockdowns imposed by the South African government to manage the COVID-19 pandemic27,28,29 has afforded us a unique opportunity to quantify shorebird responses to increasing human density in Muizenberg Beach over 8 months in 2020, including a 2-month period of virtual human exclusion. In spite of our study being limited to one beach over 2 years, we were able to take advantage of data collected prior to- (2019) and during the 2020 COVID lockdowns, to better understand a pervasive feature of sandy beach ecosystems (human recreation) that is predicted to intensify in future10.Findings for the 2019–2020 component of our study generally conformed to hypotheses posed. Firstly, shorebird abundance was inversely associated with human abundance and was positively related to lockdown level in 2020. Secondly, shorebird abundance was generally greatest during lockdown levels 5 and 4, when humans were effectively absent from the beach. To contextualise, shorebird abundance was roughly six times greater at the start of lockdown level 5 (2020) than the equivalent period in 2019. Thirdly, lowest shorebird abundance occurred during lockdown level 1 when human abundance was greatest in 2020. Collectively, these findings indicate a strong inverse association between shorebird- and human abundance on Muizenberg Beach and align with results of other studies36,37,38,39. Cumulatively, our findings, allied with prior research highlight the potential for human recreational activity, particularly at high intensities, to impact shorebird utilisation of sandy beach ecosystems, which may in turn affect ecological functions they provide that contribute to ecosystem multifunctionality.The inverse relationship that we recorded between human- and shorebird abundance likely manifests through the diverse ways in which recreational activity impacts fundamental processes and ecosystem components, which in turn link ecologically to shorebirds10,36,37,38,39,40. Muizenberg Beach is popular for surfing, bait-harvesting and general recreational activities, and it is these activities that likely drive the human-shorebird relationship that we report, particularly in 2020. When carried out under high human densities, such activities can lead to a reduction in space available, rendering the ecosystem less suitable as a substrate for birds36. Noise pollution and the presence of dogs may further depress habitat suitability41. Repeated trampling of sediment can negatively impact macrofaunal populations, which together with altered sedimentary biogeochemistry (e.g. increased anoxia), can reduce trophic resource availability to shorebirds, with benthic bait-collecting compounding these effects42,43. At the start of our data collection in 2020, we were unable to identify shorebird species due to lockdown levels 5 and 4 prohibiting human presence on the beach27,28,29. It is probable though that shorebird assemblages during lockdown levels 5 and 4 were not the same as those we identified between lockdown level 3 to 1 (mainly gulls; Table 3). This is based on research showing that increasing environmental disturbances can induce switches in biotic assemblages to those that can tolerate human activities44. Thus, the shorebird assemblages we identified during lockdown levels 3 to 1 is potentially the end-result of the mechanisms highlighted above (space reduction, noise, reduced resource availability) acting on shorebird assemblages in the absence of humans (lockdown levels 5 and 4) following humans being permitted onto the beach.At an inter-annual level, our data revealed idiosyncratic patterns that raise interesting questions about human-shorebird relationships. In 2019, in the absence of any lockdowns, shorebird abundance rose over the winter period (May–August). Winter peaks in abundance have previously been recorded in the literature45,46,47, including for kelp gulls (Larus dominicanus), which were the dominant shorebird in Muizenberg Beach. Specifically, winter abundance peaks for this species have been recorded in sandy beaches in the Eastern Cape, the Swartkops Estuary and Algoa Bay in South Africa (southeast coast)45,46,47. However, the absence of a winter abundance peak in 2020 raises the possibility that the 2019 winter-peak was not seasonal but an opportunistic response to decreased human abundance (see Fig. 4A). In South Africa, coastal ecosystems generally experience greatest human numbers in summer, due to warmer conditions and long end-of-year-vacation periods, based on our observations and experiences.The second inter-annual trend worth noting in our findings is that shorebird abundance was greater in 2019 than 2020, despite lockdowns being implemented in 2020. This counterintuitive finding is likely due to lockdowns that excluded people from the beach in 2020 (levels 5 to 3) being too short in duration to facilitate increases in bird numbers in 2020 beyond the 2019 level. This is supported by our data showing that humans were excluded from the beach for a total of 2 months (April and May 2020; levels 5-4) out of the 8-month period during which photographs were analysed. It would have been expected at the onset of the study that humans would be excluded from the beach during lockdown level 329, which would have resulted in an additional two and a half months of human exclusion and potentially a higher mean shorebird abundance for 2020. However, it is clear from our data that humans were present on the beach during level 3. On closer inspection, it is evident that human numbers increased even prior to the end of lockdown level 4. In fact, human abundance was greater under lockdown level 3 in 2020 than in the same period in 2019. Such high numbers of humans on the beach despite prohibitions are likely due to a lack of compliance, confusion around regulations and/or ‘covid fatigue’, which describes the propensity of humans to grow tired of COVID-19 regulations48. An additional consideration is that human numbers on the beach increased dramatically during lockdown levels 2 and 1, being almost twice the level recorded in 2019 in the same period. The lower 2020 bird count that we recorded is thus likely a product of the short duration of human exclusions in 2020 (lockdown levels 4 and 5) and the magnitude and rate of increase in human numbers thereafter (levels 3-1). Separately, our findings additionally suggest that surrogates (lockdown levels in our case) are unreliable estimators of human presence or abundance and align with findings elsewhere24.The last noteworthy inter-annual trend in our data was the difference in strength of human-shorebird relationships. While the inverse relationship between human and shorebird numbers was evident in both years, it was only during 2020, when humans were excluded from Muizenberg Beach, that the extent of this relationship was revealed. Specifically, in 2020, human exclusion at the start of lockdown level 5 was accompanied by a six-fold increase in shorebird abundance relative to 2019 at the same period. Additional support for the difference in strength of the human-shorebird relationship is the (1) significant interaction recorded between human numbers and year in explaining shorebird abundance and (2) the almost twofold stronger negative relationship (based on regression slopes) between shorebird and human abundance in 2020 vs 2019. These findings suggest that were it not for the COVID lockdowns in 2020, the extent of increasing human numbers on shorebirds may have been masked. However, it must be borne in mind that inter-annual variation may have played some role in the difference in trends recorded for 2019 versus 2020, though we cannot quantify this, given that we only have data for 2 years. Nevertheless, we suggest that when making conservation/management recommendations, decision-makers need to be cognisant of the potential for human effects on sandy beach ecosystems to be underestimated in studies based on variation in human density, in which human exclusion at appropriate spatial and temporal scales is absent24. Concerns have been expressed in the past about the failure of studies to consistently detect large-scale changes in sandy beach ecosystems, including those induced by recreational activities19. We suggest that such deficiencies may relate in part to the scarcity of true human exclusions in disturbance studies at meaningful scales in space and time.Findings from the in situ component of our study suggested that shorebird assemblages were negligibly affected by the transition from lockdown level 3 to 1, but that spatial differences among zones were more prominent. The lack of cases in which lockdown levels interacted statistically with zones (Tables 2, 4) further reinforces our conclusion regarding lockdown effects. Shorebird assemblage structure did vary between lockdown levels 3 and 2, due mainly to increasing contributions of Chroicocephalus hartlaubii (Hartlaub’s Gull) from level 3 to 2 and the opposite for L. dominicanus. Contrary to our hypothesis, differences in assemblage (Shannon–Wiener diversity was the exception) and species metrics were not detected among lockdown levels. This was likely due to the gradient in human abundance being weak among lockdown levels 3 to 1, relative to levels 5 and 4, with there being no virtual exclusion of humans under level 3 lockdown, as would have been expected given government regulations29. It is also possible that under lockdown levels 3, 2 and 1, the shorebird assemblage was simplified and comprised species tolerant of human activities44. The increase in Shannon–Wiener diversity value from lockdown level 3 to 2 was counter expectation, but likely reflects increased evenness during lockdown level 2, brought on by the declining dominance of L. dominicanus and a greater contribution of C. hartlaubii.Taken in its entirety, our findings provide valuable perspectives on human-shorebird interactions in sandy beaches. Based on our 2020 data spanning lockdowns of decreasing severity, our findings suggest that shorebirds are likely to benefit from human-free periods. This benefit is in reality likely to extend across multiple-trophic levels and is unlikely to be shorebird-specific, based on prior research reporting positive organism metrics at lower trophic levels in low human and/or human-free conditions in beach ecosystems20. Broadly, our findings attest to the value of using current and future lockdowns associated with managing the global COVID-19 pandemic to provide data on responses of birds and other organism groups to human-free spaces and times25,26,49. These human-free conditions can additionally provide invaluable data on sensitivities of ecosystem components and processes to increasing human density25,26,49. Data collected during lockdowns can provide better approximations of baseline conditions in sandy beach ecosystems, thereby providing a more meaningful basis for (1) evaluating future ecosystem change in response to human and global change stressors and (2) developing ecosystem restoration programs. This would be central to preventing long-term ecosystem degradation through the shifting base-line syndrome, where successive generations of decision makers/scientists judge the magnitude of change experienced by ecosystem components against increasingly deteriorating conditions over generational time-scales50. We also advocate for data emanating from COVID lockdown studies to be used in public education initiatives, so that beach users are made aware of the ways in which recreational activities can influence beach ecosystems. Such initiatives can improve involvement of public stakeholders in management of sandy beach ecosystems, which has been shown to provide cost-effective and sound decision-making, while increasing support for conservation initiatives51,52,53.Lastly, our findings have shed light on the sensitivity of shorebirds to increasing human numbers, mainly for recreational purposes. By moving beyond binary contrasts of human presence/absence, our work has also shown the magnitude of increasing human numbers on shorebirds, by virtue of the 34.18% increase in human abundance in our study corresponding with a 79.63% decline in bird numbers during the transition from lockdown level 4 to 3 in 2020. This finding is highly relevant considering that our work was based on an urban ecosystem—such systems are thought to have avian communities that are more disturbance tolerant relative to rural or suburban ecosystems54. Broadly, our work emphasises the need for environmental managers and city planners to be cognisant of the sensitivity of shorebirds to human recreational activities, even in urban settings, and to develop appropriate management plans in conjunction with scientists and stakeholders51,52,53. It should be noted that bird responses that we recorded in 2020 are unlikely to be driven solely by changing human numbers in Muizenberg Beach. Processes influencing bird assemblages in beaches surrounding our focal study area, including changes in human numbers and behaviour, may also have been influential determinants of trends recorded. We lack the data to comment meaningfully on this, but is an area worth exploring in future studies.Concluding perspectivesThe global COVID-19 anthropause has been described as the greatest large-scale experiment in modern history. This period has afforded scientists a unique opportunity to refine understanding of the consequences of human activities on Earth’s natural environments25,26,49. This is particularly relevant for human-dominated ecosystems such as sandy beaches, which are arguably the most utilised of Earth’s ecosystems for recreational purposes. In the absence of the COVID-19 anthropause, it is doubtful whether human exclusions could be carried out at scales that would allow meaningful detection of responses to human recreational disturbance. Our findings broadly attest to the points raised thus far, illustrating not only the potential for conventional approaches to underestimate human effects in sandy beaches, but also the sensitivity of shorebirds to human recreation and the magnitude of human influence. We hope that our findings stimulate further research on human recreational effects on sandy beach ecosystems, particularly with a view towards quantifying disturbance sensitivities and response thresholds of fundamental processes that drive multifunctionality in these heavily utilised, yet highly significant coastal ecosystems. We suggest that this is an imperative, given the exponential human population growth expected in the future, particularly along the coast, and the increasing demand predicted on sandy beach ecosystems from recreation, tourism and commercial sectors10,18. At its broadest level, our work dovetails with prior calls for scientists to capitalise on current and future COVID lockdowns to refine our understanding of human-nature interactions25, so that ecosystems and socio-ecological services provided can be sustainably utilised in the future. More

A biodiversity simulation frameworkWe have developed a simulation framework modelling biodiversity loss to optimize and validate conservation policies (in this context, decisions about data gathering and area protection across a landscape) using an RL algorithm. We implemented a spatially explicit individual-based simulation to assess future biodiversity changes based on natural processes of mortality, replacement and dispersal. Our framework also incorporates anthropogenic processes such as habitat modifications, selective removal of a species, rapid climate change and existing conservation efforts. The simulation can include thousands of species and millions of individuals and track population sizes and species distributions and how they are affected by anthropogenic activity and climate change (for a detailed description of the model and its parameters see Supplementary Methods and Supplementary Table 1).In our model, anthropogenic disturbance has the effect of altering the natural mortality rates on a species-specific level, which depends on the sensitivity of the species. It also affects the total number of individuals (the carrying capacity) of any species that can inhabit a spatial unit. Because sensitivity to disturbance differs among species, the relative abundance of species in each cell changes after adding disturbance and upon reaching the new equilibrium. The effect of climate change is modelled as locally affecting the mortality of individuals based on species-specific climatic tolerances. As a result, more tolerant or warmer-adapted species will tend to replace sensitive species in a warming environment, thus inducing range shifts, contraction or expansion across species depending on their climatic tolerance and dispersal ability.We use time-forward simulations of biodiversity in time and space, with increasing anthropogenic disturbance through time, to optimize conservation policies and assess their performance. Along with a representation of the natural and anthropogenic evolution of the system, our framework includes an agent (that is, the policy maker) taking two types of actions: (1) monitoring, which provides information about the current state of biodiversity of the system, and (2) protecting, which uses that information to select areas for protection from anthropogenic disturbance. The monitoring policy defines the level of detail and temporal resolution of biodiversity surveys. At a minimal level, these include species lists for each cell, whereas more detailed surveys provide counts of population size for each species. The protection policy is informed by the results of monitoring and selects protected areas in which further anthropogenic disturbance is maintained at an arbitrarily low value (Fig. 1). Because the total number of areas that can be protected is limited by a finite budget, we use an RL algorithm42 to optimize how to perform the protecting actions based on the information provided by monitoring, such that it minimizes species loss or other criteria depending on the policy.We provide a full description of the simulation system in the Supplementary Methods. In the sections below we present the optimization algorithm, describe the experiments carried out to validate our framework and demonstrate its use with an empirical dataset.Conservation planning within a reinforcement learning frameworkIn our model we use RL to optimize a conservation policy under a predefined policy objective (for example, to minimize the loss of biodiversity or maximize the extent of protected area). The CAPTAIN framework includes a space of actions, namely monitoring and protecting, that are optimized to maximize a reward R. The reward defines the optimality criterion of the simulation and can be quantified as the cumulative value of species that do not go extinct throughout the timeframe evaluated in the simulation. If the value is set equal across all species, the RL algorithm will minimize overall species extinctions. However, different definitions of value can be used to minimize loss based on evolutionary distinctiveness of species (for example, minimizing phylogenetic diversity loss), or their ecosystem or economic value. Alternatively, the reward can be set equal to the amount of protected area, in which case the RL algorithm maximizes the number of cells protected from disturbance, regardless of which species occur there. The amount of area that can be protected through the protecting action is determined by a budget Bt and by the cost of protection ({C}_{t}^{c}), which can vary across cells c and through time t.The granularity of monitoring and protecting actions is based on spatial units that may include one or more cells and which we define as the protection units. In our system, protection units are adjacent, non-overlapping areas of equal size (Fig. 1) that can be protected at a cost that cumulates the costs of all cells included in the unit.The monitoring action collects information within each protection unit about the state of the system St, which includes species abundances and geographic distribution:$${S}_{t}={{{{H}}}_{{{t}}},{{{D}}}_{{{t}}},{{{F}}}_{{{t}}},{{{T}}}_{{{t}}},{{{C}}}_{{{t}}},{{{P}}}_{{{t}}},{B}_{t}}$$
(1)
where Ht is the matrix with the number of individuals across species and cells, Dt and Ft are matrices describing anthropogenic disturbance on the system, Tt is a matrix quantifying climate, Ct is the cost matrix, Pt is the current protection matrix and Bt is the available budget (for more details see Supplementary Methods and Supplementary Table 1). We define as feature extraction the result of a function X(St), which returns for each protection unit a set of features summarizing the state of the system in the unit. The number and selection of features (Supplementary Methods and Supplementary Table 2) depends on the monitoring policy πX, which is decided a priori in the simulation. A predefined monitoring policy also determines the temporal frequency of this action throughout the simulation, for example, only at the first time step or repeated at each time step. The features extracted for each unit represent the input upon which a protecting action can take place, if the budget allows for it, following a protection policy πY. These features (listed in Supplementary Table 2) include the number of species that are not already protected in other units, the number of rare species and the cost of the unit relative to the remaining budget. Different subsets of these features are used depending on the monitoring policy and on the optimality criterion of the protection policy πY.We do not assume species-specific sensitivities to disturbance (parameters ds, fs in Supplementary Table 1 and Supplementary Methods) to be known features, because a precise estimation of these parameters in an empirical case would require targeted experiments, which we consider unfeasible across a large number of species. Instead, species-specific sensitivities can be learned from the system through the observation of changes in the relative abundances of species (x3 in Supplementary Table 2). The features tested across different policies are specified in the subsection Experiments below and in the Supplementary Methods.The protecting action selects a protection unit and resets the disturbance in the included cells to an arbitrarily low level. A protected unit is also immune from future anthropogenic disturbance increases, but protection does not prevent climate change in the unit. The model can include a buffer area along the perimeter of a protected unit, in which the level of protection is lower than in the centre, to mimic the generally negative edge effects in protected areas (for example, higher vulnerability to extreme weather). Although protecting a disturbed area theoretically allows it to return to its initial biodiversity levels, population growth and species composition of the protected area will still be controlled by the death–replacement–dispersal processes described above, as well as by the state of neighbouring areas. Thus, protecting an area that has already undergone biodiversity loss may not result in the restoration of its original biodiversity levels.The protecting action has a cost determined by the cumulative cost of all cells in the selected protection unit. The cost of protection can be set equal across all cells and constant through time. Alternatively, it can be defined as a function of the current level of anthropogenic disturbance in the cell. The cost of each protecting action is taken from a predetermined finite budget and a unit can be protected only if the remaining budget allows it.Policy definition and optimization algorithmWe frame the optimization problem as a stochastic control problem where the state of the system St evolves through time as described in the section above (see also Supplementary Methods), but it is also influenced by a set of discrete actions determined by the protection policy πY. The protection policy is a probabilistic policy: for a given set of policy parameters and an input state, the policy outputs an array of probabilities associated with all possible protecting actions. While optimizing the model, we extract actions according to the probabilities produced by the policy to make sure that we explore the space of actions. When we run experiments with a fixed policy instead, we choose the action with highest probability. The input state is transformed by the feature extraction function X(St) defined by the monitoring policy, and the features are mapped to a probability through a neural network with the architecture described below.In our simulations, we fix monitoring policy πX, thus predefining the frequency of monitoring (for example, at each time step or only at the first time step) and the amount of information produced by X(St), and we optimize πY, which determines how to best use the available budget to maximize the reward. Each action A has a cost, defined by the function Cost(A, St), which here we set to zero for the monitoring action (X) across all monitoring policies. The cost of the protecting action (Y) is instead set to the cumulative cost of all cells in the selected protection unit. In the simulations presented here, unless otherwise specified, the protection policy can only add one protected unit at each time step, if the budget allows, that is if Cost(Y, St)  More

Corwin, D. L. & Lesch, S. M. Apparent soil electrical conductivity measurements in agriculture. Comput. Electron. Agric. 46, 11–43 (2005).
Google Scholar 
Akramkhanov, A., Lamers, J. P. A. & Martius, C. Conversion factors to estimate soil salinity based on electrical conductivity for soils in Khorezm region, Uzbekistan. In Sustainable Management of Saline Waters and Salt-Affected Soils for Agriculture (ed. Qadir, S. et al.) 19–25 (Syria, 2009).Boettinger, J. L., Doolittle, J. A., West, N. E., Bork, E. W. & Schupp, E. W. Nondestructive assessment of rangeland soil depth to petrocalcic horizon using electromagnetic induction. Arid. Land Res. Manag. 11, 372–390 (1997).
Google Scholar 
Herrero, J., Ba, A. A. & Aragues, R. Soil salinity and its distribution determined by soil sampling and electromagnetic techniques. Soil Use Manag. 19, 119–126 (2003).
Google Scholar 
Corwin, D. L. Past, present, and future trends in soil electrical conductivity measurements using geophysical methods. In Handbook of Agricultural Geophysics (eds. Allred, B. J., Daniels, J. J. & Ehsani, M. R.) 17–44 (Boca Raton, 2008).Triantafillou, J., Lesch, S. M., La Lau, K. & Buchanan, S. M. Field level digital mapping of cation exchange capacity using electromagnetic induction and a hierarchical spatial regression model. Aust. J. Soil Res. 47, 651–663 (2009).
Google Scholar 
Lardo, E., Arous, A., Palese, A. M., Nuzzo, V. & Celano, G. Electromagnetic induction: A support tool for the evaluation of soil CO2 emissions and soil organic carbon content in olive orchards under semi-arid conditions. Geoderma 264, 188–194 (2016).ADS 
CAS 

Google Scholar 
Yao, R. J. et al. Geostatistical monitoring of soil salinity for precision management using proximally sensed electromagnetic induction (EMI) method. Environ. Earth Sci. 75(20), 1362. https://doi.org/10.1007/s12665-016-6179-z (2016).CAS 

Google Scholar 
Corwin, D. L. & Lesch, S. M. Protocols and guidelines for field-scale measurement of soil salinity distribution with ECa-directed soil sampling. J. Environ. Eng. Geophys. 18(1), 1–25 (2013).
Google Scholar 
Heil, K. & Schmidhalter, U. The application of EM38: Determination of soil parameters, selection of soil sampling points and use in agriculture and archaeology. Sensors. 17, 2540 (2017).ADS 
PubMed Central 

Google Scholar 
Rhoades, J. D., Corwin, D. L. & Lesch, S. M. Geospatial measurements of soil electrical conductivity to assess soil salinity and diffuse salt loading from irrigation. In Assessment of Non-point Source Pollution in the Vadose Zone (eds. Corwin, D. L., Loague, K. & Ellsworth, T. R.) 197–215 (Geophysical Monogram, 1999).Sadler, E. J., Camp, C. R. & Evans, R. G. New and future technology. In Irrigation of Agricultural Crops (eds. Steward, B. A. & Nelson, D. R.) 609–626 (Agronomy Monograph, 2007).Carrow, R. N., Krum, J. M., Flitcroft, I. & Cline, V. Precision turfgrass management: Challenges and field application for mapping turfgrass soil and stress. Precis. Agric. 11, 115–134 (2010).
Google Scholar 
Devitt, D. A., Lockett, M. & Bird, B. M. Spatial and temporal distribution of salts on fairways and greens irrigated with reuse water. Agronomy 99, 692–700 (2007).
Google Scholar 
Corwin D.L., Lesch S.M. & Lobell D.B. Laboratory and field measurements. In Agricultural Salinity Assessment and Management (eds. Wallender, W. W. & Tanji, K. K.) (2012).Lesch, S. M., Rhoades, J. D., Corwin, D. L., Robinson, D. A. & Suárez, D. L. ESAP-RSSD version 2.30R. User manual and tutorial guide. Res. Report 148 in USDA-ARS. George E. Brown, Jr., Salinity Laboratory, Riverside, California. (2002).Lesch, S. M., Rhoades, J. D., Corwin, D. L., Robinson, D. A. & Suárez, D. L. ESAP-SaltMapper version 2.30R. User manual and tutorial guide. Res. Report 149 USDA-ARS. George E. Brown, Jr., Salinity Laboratory, Riverside, California. (2002).Lesch, S. M., Rhoades, J. D. & Corwin, D. L. ESAP-95 Version 2.01R: User manual and tutorial guide. Res. Rep. 146. USDA-ARS. George E. Brown, Jr., Salinity Laboratory, Riverside, California. (2000).Lesch, S. M., Strauss, D. J. & Rhoades, J. D. Spatial prediction of soil salinity using electromagnetic induction techniques: 1. Statistical prediction models: A comparison of multiple linear regression and cokriging. Water Resour. Res. 31, 373–386 (1995).ADS 

Google Scholar 
Amezketa, E. Soil salinity assessment using directed soil sampling from a geophysical survey with electromagnetic technology: A case study. Span. J. Agric. Res. 5(1), 91–101 (2007).
Google Scholar 
Grieve C. M., Grattan, S. R. & Mass, E. V. Plant salt tolerance. In Agricultural Salinity Assessment and Management (eds. Walender W. W. & Tanji K.K.) (ASCE, 2012).Shahba, M. Interaction effects of salinity and mowing on performance and physiology of bermudagrass cultivars. Crop Sci. 50, 2620–2631 (2010).
Google Scholar 
Marcum, K. B. & Pessarakli, M. Salinity tolerance and salt gland excretion efficiency of bermudagrass turf cultivars. Crop Sci. 46, 2571–2574 (2006).
Google Scholar 
Xiang, M., Moss, J. Q., Martin, D. L., Su, K. & Dunn, B. L. Evaluating the salinity tolerance of clonal-type bermudagrass cultivars and an experimental selection. Hortic. Sci. 51(1), 185–191 (2017).
Google Scholar 
Ganjegunte, G. K. et al. Soil salinity of an urban park after long term irrigation with saline ground water. Agronomy 109, 3011–3018 (2017).CAS 

Google Scholar 
Keren, R. & Miyamoto, S. Reclamation of saline, sodic and boron affected soils. In Agricultural Salinity Assessment and Management (eds. Walender W. W. & Tanji K. K.) (ASCE, 2012).Thomas, G. W. & Phillips, R. E. Consequences of water-movement in macropores. J. Environ. Qual. 8, 149–152 (1979).
Google Scholar 
White, R. E. The influence of macropores on the transport of dissolved and suspended matter through soil. In Advances in Soil Science (ed. Stewart, B. A.) 95–120 (Springer, 1985).Workman, S. & Skaggs, R. PREFLO: A water management model capable of simulating preferential flow. Trans. ASAE. 33, 1939–1948 (1990).
Google Scholar 
Huang, B., Duncan, R. R. & Carrow, R. N. Drought-resistance mechanisms of seven warm-season turfgrasses under surface soil drying: II. Root aspects. Crop Sci. 7(6), 863–1869 (1997).
Google Scholar 
Liu, X. H. & Huang, B. R. Cytokinin effects on creeping bentgrass response to heat stress: II. Leaf senescence and antioxidant metabolism. Crop Sci. 42, 466–472 (2002).CAS 

Google Scholar  More