Philippa Kaur

Alvares, F. et al. Old Wolrd Canis spp. with taxonomic ambiguity: Workshop conclusions and recommendations Vairao, Portugal, 28th–30th May 2019. Canid News (Online Edition) (2019).Jackson, S. M. et al. Taxonomy of the dingo: It’s an ancient dog. Aust. Zool. 41, 347–357 (2021).
Google Scholar 
Stephens, D., Wilton, A. N., Fleming, P. J. S. & Berry, O. Death by sex in an Australian icon: A continent-wide survey reveals extensive hybridization between dingoes and domestic dogs. Mol. Ecol. 24, 5643–5656 (2015).CAS 
PubMed 

Google Scholar 
Cairns, K. M., Shannon, L. M., Koler-Matznick, J., Ballard, J. W. O. & Boyko, A. R. Elucidating biogeographical patterns in Australian native canids using genome wide SNPs. PLoS ONE 13, e0198754 (2018).PubMed 
PubMed Central 

Google Scholar 
Fleming, P. J. S., Ballard, G. & Cutter, N. There is no Dingo dilemma: legislation facilitates culling, containment and conservation of Dingoes in New South Wales. Aust. Zool. 41, 408–416 (2021).
Google Scholar 
Corbett, L. K. The Dingo in Australia and Asia. Second edn, (JB Books Australia, 2001).Newsome, T. M. et al. Making a new dog?. Bioscience 67, 374–381 (2017).
Google Scholar 
Wang, G.-D. et al. Out of southern East Asia: the natural history of domestic dogs across the world. Cell Res. 26, 21–33 (2016).PubMed 

Google Scholar 
Smith, B. The Dingo Debate: Origins, Behaviour and Conservation. (CSIRO Publishing, 2015).Jackson, S. M. et al. The dogma of dingoes-taxonomic status of the dingo: A reply to Smith et al. Zootaxa 4564, 198–212 (2019).
Google Scholar 
Zhang, S. J. et al. Genomic regions under selection in the feralization of the dingoes. Nat. Commun. 11, 671 (2020).ADS 
PubMed 
PubMed Central 

Google Scholar 
Balme, J. & O’Connor, S. Dingoes and Aboriginal social organization in Holocene Australia. J. Archaeol. Sci. Rep. 7, 775–781 (2016).
Google Scholar 
Cairns, K. M. What is a dingo – origins, hybridisation and identity. Aust. Zool. 41(3), 322–337 (2021).
Google Scholar 
Allen, B. L. & West, P. Influence of dingoes on sheep distribution in Australia. Aust. Vet. J. 91, 261–267 (2013).CAS 
PubMed 

Google Scholar 
Fleming, P. J. S. in Carnivores of Australia: Past, Present and Future (eds A.S. Glen & C.R. Dickman) Ch. 6, 105–149 (CSIRO Publishing, 2014).Stephens, D. The molecular ecology of Australian wild dogs: hybridisation, gene flow and genetic structure at multiple geographic scales, The University of Western Australia, (2011).Cairns, K. M., Nesbitt, B. J., Laffan, S. W., Letnic, M. & Crowther, M. S. Geographic hot spots of dingo genetic ancestry in southeastern Australia despite hybridisation with domestic dogs. Conserv. Genet. 21, 77–90 (2020).CAS 

Google Scholar 
Wilton, A. N., Steward, D. J. & Zafiris, K. Microsatellite variation in the Australian dingo. J. Hered. 90, 108–111 (1999).CAS 
PubMed 

Google Scholar 
Allendorf, F. W., Leary, R. F., Spruell, P. & Wenburg, J. K. The problems with hybrids: Setting conservation guidelines. Trends Ecol. Evol. 16, 613–622 (2001).
Google Scholar 
Atkinson, J. An account of the state of agriculture & grazing in New South Wales. (J. Cross, 1826).Massy, C. The Australian Merino: The Story of a Nation (Revised and updated). xxii,1262 (Random House Australia, 2007).Cairns, K. M., Brown, S. K., Sacks, B. N. & Ballard, J. W. O. Conservation implications for dingoes from the maternal and paternal genome: Multiple populations, dog introgression, and demography. Ecol. Evol. 7, 9787–9807 (2017).PubMed 
PubMed Central 

Google Scholar 
Kopelman, N. M., Mayzel, J., Jakobsson, M., Rosenberg, N. A. & Mayrose, I. Clumpak: A program for identifying clustering modes and packaging population structure inferences across K. Mol. Ecol. Resour. 15, 1179–1191 (2015).CAS 
PubMed 
PubMed Central 

Google Scholar 
Driscoll, C., Yamaguchi, N., O’Brien, S. J. & Macdonald, D. W. A suite of genetic markers useful in assessing wildcat (Felis silvestris ssp.)-domestic cat (Felis silvestris catus) admixture. J. Hered. 102(1), S87–S90 (2011).PubMed 
PubMed Central 

Google Scholar 
Bohling, J. H. & Waits, L. P. Factors influencing red wolf–coyote hybridization in eastern North Carolina USA. Biol. Conserv. 184, 108–116 (2015).
Google Scholar 
Fleming, P., Corbett, L., Harden, R. & Thomson, P. in Managing the Impacts of Dingoes and Other Wild Dogs. (Bureau of Rural Sciences, Canberra, 2001).Van Veldhuisen, R. Pipe dreams: A history of water supply in the Wimmera-Mallee (Wimmera Mallee Water, 2001).Newsome, A. The distribution of red kangaroos, Megaleia rufa (Desmarest), about sources of persistent food and water in central Australia. Aust. J. Zool. 13, 289–300 (1965).
Google Scholar 
James, C. D., Landsberg, J. & Morton, S. R. Provision of watering points in the Australian arid zone: A review of effects on biota. J. Arid Environ. 41, 87–121 (1999).ADS 

Google Scholar 
Robinson, J. A., Brown, C., Kim, B. Y., Lohmueller, K. E. & Wayne, R. K. Purging of strongly deleterious mutations explains long-term persistence and absence of inbreeding depression in island foxes. Curr. Biol. 28, 3487-3494.e3484 (2018).CAS 
PubMed 
PubMed Central 

Google Scholar 
Benazzo, A. et al. Survival and divergence in a small group: The extraordinary genomic history of the endangered Apennine brown bear stragglers. PNAS 114, E9589–E9597 (2017).CAS 
PubMed 
PubMed Central 

Google Scholar 
Mattucci, F. et al. Genomic approaches to identify hybrids and estimate admixture times in European wildcat populations. Sci. Rep. 9, 11612 (2019).ADS 
PubMed 
PubMed Central 

Google Scholar 
Thomson, P. C., Rose, K. & Kok, N. E. The behavioural ecology of dingoes in north-western Australia. VI. Temporary extra-terrestrial movements and dispersal. Wildl. Res. 19, 585–595 (1992).
Google Scholar 
Newsome, T. M., Ballard, G.-A., Dickman, C. R., Fleming, P. J. S. & van de Ven, R. Home range, activity and sociality of a top predator, the dingo: A test of the Resource Dispersion Hypothesis. Ecography 36, 914–925 (2013).
Google Scholar 
Giglio, R. M., Rocke, T. E., Osorio, J. E. & Latch, E. K. Characterizing patterns of genomic variation in the threatened Utah prairie dog: Implications for conservation and management. Evol. Appl. 14, 1036–1051 (2021).PubMed 

Google Scholar 
Conroy, G. C. et al. Conservation concerns associated with low genetic diversity for K’gari–Fraser Island dingoes. Sci. Rep. 11, 9503 (2021).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Frankham, R. Do island populations have less genetic variation than mainland populations?. Heredity 78, 311–327 (1997).PubMed 

Google Scholar 
Funk, W. C. et al. Adaptive divergence despite strong genetic drift: genomic analysis of the evolutionary mechanisms causing genetic differentiation in the island fox (Urocyon littoralis). Mol. Ecol. 25, 2176–2194 (2016).CAS 
PubMed 
PubMed Central 

Google Scholar 
Behrendorff, L. Best-practice dingo management: six lessons from K’gari (Fraser Island). Aust. Zool. 41, 521–533 (2021).
Google Scholar 
van Eeden, L. M., Smith, B. P., Crowther, M. S., Dickman, C. R. & Newsome, T. M. ‘The dingo menace’: An historic survey on graziers’ management of an Australian carnivore. Pac. Conserv. Biol. 25, 245–256 (2019).
Google Scholar 
Whiting, S. D., Long, J. L., Hadden, K. M., Lauder, A. D. K. & Koch, A. U. Insights into size, seasonality and biology of a nesting population of the Olive Ridley turtle in northern Australia. Wildl. Res. 34, 200–210 (2007).
Google Scholar 
Banks, S. C., Hoyle, S. D., Horsup, A., Sunnucks, P. & Taylor, A. C. Demographic monitoring of an entire species (the northern hairy-nosed wombat, Lasiorhinus krefftii) by genetic analysis of non-invasively collected material. Anim. Conserv. 6, 101–107 (2003).
Google Scholar 
Parker, H. G. et al. Genomic analyses reveal the influence of geographic origin, migration, and hybridization on modern dog breed development. Cell Rep. 19, 697–708 (2017).CAS 
PubMed 
PubMed Central 

Google Scholar 
Thackway, R. & Cresswell, I. An Interim Biogeographic Regionalisation for Australia: A Framework for Setting Priorities in the National Reserves System Cooperative Program. Version 4, (Australian Nature Conservation Agency, Reserve Systems Unit, 1995).Bureau of Meteorology & CSIRO. (Bureau of Meteorology, CSIRO and Farmlink, http://www.bom.gov.au/climate/climate-guides/guides/01-Mallee-VIC-Climate-Guide.pdf, 2019).Rowan, J. N. & Downes, R. G. in Soil Conservation Authority of Victoria (ed Brookes, A.C.) 1–55 (Govt. Printer, Melbourne, 1963).Longmire, J. L., Maltbie, M. & Baker, R. J. Use of “lysis buffer” in DNA isolation and its implications for museum collections. Occas. Pap. Mus. Tex. Tech. Univ. 163, 1–3 (1997).
Google Scholar 
Tatler, J., Prowse, T. A. A., Roshier, D. A., Cairns, K. M. & Cassey, P. Phenotypic variation and promiscuity in a wild population of pure dingoes (Canis dingo). J. Zool. Syst. Evol. Res. 59, 311–322 (2020).
Google Scholar 
Pritchard, J. K., Stephens, M. & Donnelly, P. Inference of population structure using multilocus genotype data. Genetics 155, 945–959 (2000).CAS 
PubMed 
PubMed Central 

Google Scholar 
Francis, R. M. pophelper: An R package and web app to analyse and visualize population structure. Mol. Ecol. Resour. 17, 27–32 (2017).CAS 
PubMed 

Google Scholar 
Wang, J. The computer program structure for assigning individuals to populations: Easy to use but easier to misuse. Mol. Ecol. Resour. 17, 981–990 (2017).CAS 
PubMed 

Google Scholar 
Evanno, G., Regnaut, S. & Goudet, J. Detecting the number of clusters of individuals using the software structure: A simulation study. Mol. Ecol. 14, 2611–2620 (2005).CAS 
PubMed 

Google Scholar 
Earl, D. A. & von Holdt, B. M. Structure harvester: A website and program for visualizing structure output and implementing the Evanno method. Conserv. Genet. Resour. 4, 359–361 (2012).
Google Scholar 
Verity, R. & Nichols, R. A. Estimating the number of subpopulations (K) in structured populations. Genetics 203, 1827–1839 (2016).PubMed 
PubMed Central 

Google Scholar 
Jombart, T. adegenet: A R package for the multivariate analysis of genetic markers. Bioinformatics 24, 1403–1405 (2008).CAS 
PubMed 

Google Scholar 
Jombart, T., Devillard, S. & Balloux, F. Discriminant analysis of principal components: A new method for the analysis of genetically structured populations. BMC Genet. 11, 94 (2010).PubMed 
PubMed Central 

Google Scholar 
Keenan, K. et al. diveRsity: An R package for the estimation and exploration of population genetics parameters and their associated errors. Methods Ecol. Evol. 4, 782–788 (2013).
Google Scholar 
Peakall, R. & Smouse, P. E. genalex 6: genetic analysis in Excel. Population genetic software for teaching and research. Mol. Ecol. Notes 6, 288–295 (2006).
Google Scholar 
Peakall, R. & Smouse, P. E. GenAlEx 6.5: Genetic analysis in Excel. Population genetic software for teaching and research—an update. Bioinformatics 28, 2537–2539 (2012).CAS 
PubMed 
PubMed Central 

Google Scholar 
Lynch, M. & Ritland, K. Estimation of pairwise relatedness with molecular markers. Genetics 152, 1753–1766 (1999).CAS 
PubMed 
PubMed Central 

Google Scholar 
Weir, B. S. & Cockerham, C. C. Estimating F-statistics for the analysis of population structure. Evolution 38, 1358–1370 (1984).CAS 
PubMed 

Google Scholar 
Jost, L. GST and its relatives do not measure differentiation. Mol. Ecol. 17, 4015–4026 (2008).PubMed 

Google Scholar 
Do, C. et al. NeEstimator v2: Re-implementation of software for the estimation of contemporary effective population size (Ne) from genetic data. Mol. Ecol. Resour. 14, 209–214 (2014).CAS 
PubMed 

Google Scholar 
Shirk, A. J. & Cushman, S. A. sGD: Software for estimating spatially explicit indices of genetic diversity. Mol. Ecol. Resour. 11, 922–934 (2011).CAS 
PubMed 

Google Scholar 
Schnute, J., Boers, N., Haigh, R. & Couture-Beil, A. Introduction to PBSmapping. (2016). More

Wang, S., Liu, Q. & Zhang, D. Karst rocky desertification in southwestern China: Geomorphology, landuse, impact and rehabilitation. Land Degrad. Dev. 15(2), 115–121 (2004).
Google Scholar 
Jiang, M. et al. Geologic factors leadingly drawing the macroecological pattern of rocky desertification in southwest China. Sci. Total Environ. 458–460, 419–426 (2013).
Google Scholar 
Jiang, Z., Lian, Y. & Qin, X. Rocky desertification in Southwest China: Impacts, causes, and restoration. Earth-Sci. Rev. 132, 1–12 (2014).ADS 

Google Scholar 
Xu, E., Zhang, H. & Li, M. Object-based mapping of karst rocky desertification using a support vector machine. Land Degrad. Dev. 26(2), 158–167 (2012).
Google Scholar 
Li, Y., Bai, X., Wang, S. & Tian, Y. Integrating mitigation measures for karst rocky desertification land in the Southwest mountains of China. Carbonates Evaporites 34, 1095–1106 (2018).
Google Scholar 
Lan, J. Responses of soil organic carbon components and their sensitivity to karst rocky desertification control measures in Southwest China. J. Soil. Sediment. 21, 978–989 (2020).
Google Scholar 
Gao, J., Du, F., Zuo, L. & Jiang, Y. Integrating ecosystem services and rocky desertification into identification of karst ecological security pattern. Landscape Ecol. 36, 2113–2133 (2020).
Google Scholar 
Huang, X. et al. Driving factors and prediction of rock desertification of non-tillage lands in a karst basin, Southwest China. Pol. J. Environ. Stud. 30(4), 3627–3635 (2021).CAS 

Google Scholar 
Chen, S., Zhou, Z., Yan, L. & Li, B. Quantitative evaluation of ecosystem health in a karst area of South China. Sustain. Basel 8(10), 975 (2016).
Google Scholar 
Liu, F., He, B. Y. & Kou, J. F. Landsat thermal remote sensing to investigate the present situation and variation characteristics of karst rocky desertification in Pingguo County of Guangxi, Southwest China. Sci. Soil Water Conserv. 15(02), 125–131 (2017).
Google Scholar 
Zhang, X., Shang, K., Cen, Y., Shuai, T. & Sun, Y. Estimating ecological indicators of karst rocky desertification by linear spectral unmixing method. Int. J. Appl. Earth Obs. Geoinf. 31, 86–94 (2014).ADS 
CAS 

Google Scholar 
Zhang, Z., Ouyang, Z., Xiao, Y., Xiao, Y. & Xu, W. Using principal component analysis and annual seasonal trend analysis to assess karst rocky desertification in southwestern China. Environ. Monit. Assess. 189(6), 1–19 (2017).
Google Scholar 
Li, S. & Wu, H. Mapping karst rocky desertification using Landsat 8 images. Remote Sens. Lett. 6(9), 657–666 (2015).
Google Scholar 
Yang, S. X., Lin, H., Hou, F., Zhang, L. P. & Hu, Z. L. Estimating karst area vegetation coverage by pixel unmixing. Bull. Surv. Mapp. 5, 23–27 (2014).
Google Scholar 
Xiong, Y., Yue, Y. M. & Wang, K. L. Comparative study of indicator extraction for assessment of karst rocky desertification based on hyperion and ASTER images. Bull. Soil Water Conserv. 33(03), 186–190 (2013).
Google Scholar 
Dai, G., Sun, H., Wang, B., Huang, C., Wang, W., Yao, Y., et al. Assessment of karst rocky desertification from the local to regional scale based on unmanned aerial vehicle images: Acase-study of Shilin County, Yunnan Province, China. Land Degrad. Dev. 1–14 (2021).Pu, J., Zhao, X., Dong, P., Wang, Q. & Yue, Q. Extracting information on rocky desertification from satellite images: A comparative study. Remote Sens. 13(13), 2497 (2021).ADS 

Google Scholar 
Yue, Y. M. et al. Remote sensing of indicators for evaluating karst rocky desertification. Procedia Environ. Sci. 15(04), 722–736 (2011).
Google Scholar 
Huang, Q. & Cai, Y. Spatial pattern of Karst rock desertification in the middle of Guizhou Province. Southwestern China. Environ. Geol. 52(7), 1325–1330 (2006).MathSciNet 

Google Scholar 
Wang, J., Li, S., Li, H., Luo, H. & Wang, M. Classifying indices and remote sensing image characters of rocky desertification lands: a case of karst region in Northern Guangdong Province. J. Desert Res. 5, 765–770 (2007).
Google Scholar 
Chen, F. et al. Assessing spatial-temporal evolution processes and driving forces of karst rocky desertification. Geocarto Int. 1–22 (2019).Qi, X., Zhang, C. & Wang, K. Comparing remote sensing methods for monitoring karst rocky desertification at sub-pixel scales in a highly heterogeneous karst region. Sci. Rep-UK https://doi.org/10.1038/s41598-019-49730-9 (2019).Article 

Google Scholar 
Yue, Y. et al. Spectral indices for estimating ecological indicators of karst rocky desertification. Int. J. Remote Sens. 31(8), 2115–2122 (2010).
Google Scholar 
Yan, Y., Hu, B. Q., Han, Q. Y. & Li, Y. L. Early warning for karst rocky desertification in agricultural land base on the 3S and ANN technique: A case study in Du’an County, Guangxi. Carsologica Sin. 31(01), 52–58 (2012).
Google Scholar 
Zhang, J. et al. Spectral analysis of seasonal rock and vegetation changes for detecting karst rocky desertification in southwest China. Int. J. Appl. Earth Obs. Geoinf. https://doi.org/10.1016/j.jag.2021.102337 (2021).Article 
PubMed 
PubMed Central 

Google Scholar 
Liu, Y., Wang, J. & Deng, X. Rocky land desertification and its driving forces in the karst areas of rural Guangxi, Southwest China. J. Mt. Sci-Engl. 5(4), 350–357 (2008).
Google Scholar 
Li, Y., Xie, J., Luo, G., Yang, H. & Wang, S. The evolution of a karst rocky desertification land ecosystem and its driving forces in the Houzhaihe Area, China. J. Ecol. 5, 501–512 (2015).
Google Scholar 
Zhang, Y. R., Zhou, Z. F. & Ma, S. B. Rocky desertification and climate change characteristics in typical karst area of Guizhou Province over past two decades. Environ. Sci. Technol. 37(09), 192–197 (2014).
Google Scholar 
Bai, X. Y., Wang, S. J., Chen, Q. W. & Cheng, A. Y. Constrains of lithological background of carbonate rock on spatio-temporal evolution of karst rocky desertification land. Earth Sci. 35(4), 691–696 (2010).
Google Scholar 
Li, L. & Xiong, K. Study on peak-cluster-depression rocky desertification landscape evolution and human activity-influence in South of China. Eur. J. Remote Sens. 1–9 (2020).Yao, Y. H., Shuo, N. D. Z., Zhang, J. Y., Hu, Y. F. & Kou, Z. X. Spatiotemporal characteristics of karst rocky desertification and the impact of human activities from 2010 to 2015 in Guanling County, Guizhou Province. Prog. Geogr. 38(11), 1759–1769 (2019).
Google Scholar 
Shi, K., Yang, Q. & Li, Y. Are karst rocky desertification areas affected by increasing human activity in Southern China? An empirical analysis from nighttime light data. Int. J. Environ. Res. Public Health. 16(21), 4175 (2019).PubMed Central 

Google Scholar 
Luo, X. L. et al. Analysis on the spatio- temporal evolution process of rocky desertification in Southwest Karst area. Acta Ecol. Sin. 41(02), 680–693 (2021).
Google Scholar 
Yang, Q., Jiang, Z., Yuan, D., Ma, Z. & Xie, Y. Temporal and spatial changes of karst rocky desertification in ecological reconstruction region of Southwest China. Envirov. Earth Sci. 72(11), 4483–4489 (2014).
Google Scholar 
Zhang, C., Qi, X., Wang, K., Zhang, M. & Yue, Y. The application of geospatial techniques in monitoring karst vegetation recovery in southwest China. Prog. Phys. Geog. 41(4), 450–477 (2017).
Google Scholar 
Ying, B., Xiao, S., Xiong, K., Cheng, Q. & Luo, J. Comparative studies of the distribution characteristics of rocky desertification and land use/land cover classes in typical areas of Guizhou province, China. Envirov. Earth Sci. 71(2), 631–645 (2013).
Google Scholar 
Luo, X. et al. Analysis on the spatio-temporal evolution process of rocky desertification in Southwest Karst area. Acta Ecol. Sin. 41(2), 680–693 (2021).
Google Scholar 
Chong, G. et al. Characteristics of changes in karst rocky desertification in southtern and western china and driving mechanisms. Chin. Geogr. Sci. 31, 1082–1096 (2021).
Google Scholar 
Guo, B. et al. A novel-optimal monitoring model of rocky desertification based on feature space models with typical surface parameters derived from LANDSAT_8 OLI. Degrad. Dev. 32(17), 5023–5036 (2021).
Google Scholar 
Chen, F. et al. Spatio-temporal evolution and future scenario prediction of karst rocky desertification based on CA–Markov model. Arab. J. Geosci. 14, 1262 (2021).
Google Scholar 
Wu, X., Liu, H., Huang, X. & Zhou, T. Human driving forces: Analysis of rocky desertification in karst region in Guanling County, Guizhou Province. Chin. Geogr. Sci. 21(5), 600–608 (2011).
Google Scholar 
Chen, H. et al. The evolution of rocky desertification and its response to land use changes in Wanshan Karst area, Tongren City, Guizhou Province, China. J. Agr. Resour. Environ. 37(01), 24–35 (2020).
Google Scholar 
Zerrouki, N., Dairi, A., Harrou, F., Zerrouki, Y. & Sun, Y. Efficient land desertification detection using a deep learning-driven generative adversarial network approach: A case study. Concurr. Comp-Pract. E. https://doi.org/10.1002/cpe.6604 (2021).Article 

Google Scholar 
Keskin, H., Grunwald, S. & Harris, W. Digital mapping of soil carbon fractions with machine learning. Geoderma 339, 40–58 (2019).ADS 
CAS 

Google Scholar 
Tian, Y. et al. Aboveground mangrove biomass estimation in Beibu Gulf using machine learning and UAV remote sensing. Sci. Total Environ. https://doi.org/10.1016/j.scitotenv.2021.146816 (2021).Article 
PubMed 
PubMed Central 

Google Scholar 
Xi, H. et al. Spatio-temporal characteristics of rocky desertification in typical Karst areas of Southwest China: A case study of Puding county, Guizhou province. Acta Ecol. Sin. 38(24), 8919–8933 (2018).
Google Scholar 
Deng, Y. et al. Relationship among land surface temperature and LUCC, NDVI in typical karst area. Sci. Rep-UK. 296–306 (2018).Li, S. M., Yu, L. W., Gan, S. & Yang, Y. M. Study on inversion relationship between vegetation lndex and leaf area index of rocky desertification area in southeast Yunnan based on ETM+. J. Kunming Univ. Sci. Technol. (Natl Sci.) 40(06), 31–36 (2015).
Google Scholar 
Yan, X. & Cai, Y. Multi-Scale anthropogenic driving forces of karst rocky desertification in Southwest China. Land Degrad. Dev. 26(2), 193–200 (2013).
Google Scholar 
Meyer, H., Reudenbach, C., Wollauer, S. & Nauss, T. Importance of spatial predictor variable selection in machine learning applications – Moving from data reproduction to spatial prediction. Ecol. Model. https://doi.org/10.1016/j.ecolmodel.2019.108815 (2019).Article 

Google Scholar 
Cracknell, M. & Reading, A. Geological mapping using remote sensing data: A comparison of five machine learning algorithms, their response to variations in the spatial distribution of training data and the use of explicit spatial information. Comput. Geosci-UK 63, 22–33 (2014).
ADS 

Google Scholar 
Feng, K. et al. Monitoring desertification using machine-learning techniques with multiple indicators derived from MODIS images in Mu Us Sandy Land, China. Remote Sens. 14, 2663. https://doi.org/10.3390/rs14112663 (2022).Article 
ADS 

Google Scholar 
Belgiu, M. & Drăguţ, L. Random Forest in remote sensing: A review of applications and future directions. ISPRS J. Photogramm 114, 24–31 (2016).
Google Scholar 
Chutia, D., Bhattacharyya, D. K., Sarma, K. K., Kalita, R. & Sudhakar, S. Hyperspectral remote sensing classifications: A perspective survey. Trans. GIS https://doi.org/10.1111/tgis.12164 (2015).Article 

Google Scholar 
Song, T. Q., Peng, W. X., Du, H., Wang, K. & Zeng, F. Occurrence spatial-temporal dynamics and regulation strategies of karst rocky desertification in southwest China. Acta Ecol. Sin. 34(18), 5328–5341 (2014).
Google Scholar 
Zhu, L.F. Study on the Spatial-Temporal Variation of Vegetation Coverage and Karst Rocky Desertification based on MODIS Data. Ph.D. Dissertation, Southwestern University. Chongqing, China (2018).Yang, Q. et al. Spatio-temporal evolution of rocky desertification and its driving forces in karst areas of Northwestern Guangxi, China. Environ. Earth Sci. 64, 383–393 (2011).
Google Scholar 
Mishra, N. & Chaudhuri, G. Spatio-temporal analysis of trends in seasonal vegetation productivity across Uttarakhand, Indian Himalayas, 2000–2014. Appl. Geogr. 56, 29–41 (2015).
Google Scholar 
Zhang, X., Shang, K., Cen, Y., Shuai, T. & Sun, Y. Estimating ecological indicators of karst rocky desertification by linear spectral unmixing method. Int. J. Appl. Earth. Obs. 31, 86–94 (2014).CAS 

Google Scholar 
Reshef, D. et al. Detecting novel associations in large data sets. Science 334(6062), 1518–1524 (2011).ADS 
CAS 
PubMed 
PubMed Central 
MATH 

Google Scholar 
Li, W. et al. Concentration estimation of dissolved oxygen in Pearl River Basin using input variable selection and machine learning techniques. Sci. Total Environ. https://doi.org/10.1016/j.scitotenv.2020.139099 (2021).Article 
PubMed 
PubMed Central 

Google Scholar 
Abdelhakim, A., El, H., Luis, E., Salah, E. & Abdelghani, C. Retrieving crop albedo based on radar sentinel-1 and random forest. Approach. Remote Sens. 13(16), 3181 (2021).ADS 

Google Scholar 
Rodriguez-Galiano, V. F., Ghimire, B., Rogan, J., Chica-Olmo, M. & Rigol-Sanchez, J. P. An assessment of the effectiveness of a random forest classifier for land-cover classification. ISPRS J. Photogramm 67, 93–104 (2012).
Google Scholar 
Dharumarajan, S., Bishop, T., Hegde, R. & Singh, S. Desertification vulnerability index-an effective approach to assess desertification processes: A case study in Anantapur District, Andhra Pradesh, India. Land Degrad. Dev. 29(1), 150–161 (2017).
Google Scholar 
Li, P. et al. Dynamic monitoring of desertification in ningdong based on landsat images and machine learning. Sustainability 14, 7470. https://doi.org/10.3390/su14127470 (2022).Article 

Google Scholar 
Pacheco, A. D. P., Junior, J. A. D. S., Ruiz-Armenteros, A. M. & Henriques, R. F. F. Assessment of k-nearest neighbor and random forest classifiers for mapping forest fire areas in Central Portugal using landsat-8, sentinel-2, and terra imagery. Remote Sens. 13, 1345. https://doi.org/10.3390/rs13071345 (2021).Article 
ADS 

Google Scholar  More

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

Wesche, K. et al. The Palaearctic steppe biome: A new synthesis. Biodivers. Conserv. 25, 2197–2231 (2016).
Google Scholar 
Hurka, H. et al. The Eurasian steppe belt: Status quo, origin and evolutionary history. Turczaninowia 22, 5–71 (2019).
Google Scholar 
Willner, W. et al. Formalized classification of semi-dry grasslands in central and eastern Europe. Preslia 91, 25–49 (2019).
Google Scholar 
Willis, K. J. & McElwain, J. C. The Evolution Of Plants (Oxford University Press, Oxford, 2002).Suc, J.-P. et al. Reconstruction of Mediterranean flora, vegetation and climate for the last 23 million years based on an extensive pollen dataset. Ecol. Mediterr. 44, 53–85 (2018).
Google Scholar 
Strani, F. Impact of Early and Middle Pleistocene major climatic events on the palaeoecology of Southern European ungulates. Hist. Biol. 33, 2260–2275 (2021).
Google Scholar 
Chytrý, M. et al. A modern analogue of the Pleistocene steppe-tundra ecosystem in southern Siberia. Boreas 48, 36–56 (2019).
Google Scholar 
Manafzadeh, S., Staedler, Y. M. & Conti, E. Visions of the past and dreams of the future in the orient: The Irano–Turanian region from classical botany to evolutionary studies. Biol. Rev. 92, 1365–1388 (2017).PubMed 

Google Scholar 
Seregin, A. P., Anačkov, G. & Friesen, N. Molecular and morphological revision of the Allium saxatile group (Amaryllidaceae): Geographical isolation as the driving force of underestimated speciation. Bot. J. Linn. Soc. 178, 67–101 (2015).
Google Scholar 
Friesen, N. et al. Dated phylogenies and historical biogeography of Dontostemon and Clausia (Brassicaceae) mirror the palaeogeographical history of the Eurasian steppe. J. Biogeogr. 43, 738–749 (2016).
Google Scholar 
Seidl, A. et al. Phylogeny and biogeography of the Pleistocene Holarctic steppe and semi-desert goosefoot plant. Krascheninnikovia ceratoides. Flora 262, 151504 (2020).
Google Scholar 
Seidl, A. et al. The phylogeographic history of Krascheninnikovia reflects the development of dry steppes and semi-deserts in Eurasia. Sci. Rep. 11, 6645 (2021).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Žerdoner Čalasan, A., Seregin, A. P., Hurka, H., Hofford, N. P. & Neuffer, B. The Eurasian steppe belt in time and space: Phylogeny and historical biogeography of the false flax (Camelina Crantz, Camelineae, Brassicaceae). Flora 260, 151477 (2019).
Google Scholar 
Žerdoner Čalasan, A. et al. Pleistocene dynamics of the Eurasian steppe as a driving force of evolution: Phylogenetic history of the genus Capsella (Brassicaceae). Ecol. Evol. 11, 12697–12713 (2021).PubMed 
PubMed Central 

Google Scholar 
Žerdoner Čalasan, A., German, D. A., Hurka, H. & Neuffer, B. A story from the Miocene: Clock-dated phylogeny of Sisymbrium L. (Sisymbrieae, Brassicaceae). Ecol. Evol. 11, 2573–2595 (2021).PubMed 
PubMed Central 

Google Scholar 
Zaveská, E. et al. Multiple auto- and allopolyploidisations marked the Pleistocene history of the widespread Eurasian steppe plant Astragalus onobrychis (Fabaceae). Mol. Phylogenet. Evol. 139, 106572 (2019).PubMed 

Google Scholar 
Franzke, A. et al. Molecular signals for Late Tertiary/Early Quaternary range splits of an Eurasian steppe plant: Clausia aprica (Brassicaceae). Mol. Ecol. 13, 2789–2795 (2004).CAS 
PubMed 

Google Scholar 
Friesen, N. et al. Evolutionary history of the Eurasian steppe plant Schivereckia podolica (Brassicaceae) and its close relatives. Flora 268, 151602 (2020).
Google Scholar 
Walter, H. & Straka, H. Arealkunde (Floristisch-Historische Geobotanik), second edition (Ulmer, 1970).Zimmermann, W. 50c. Familie Ranunculáceae in Gustav Hegi, Illustrierte Flora Von Mitteleuropa, Band III, Teil 3, second edition (eds. Rechinger, K. H. & Damboldt, J.) 53–341 (Parey, 1965–1974).Meusel, H., Jäger, E. & Weinert, E. Vergleichende Chorologie Der Zentraleuropäischen Flora, Band I (Gustav Fischer, 1965).Hoffmann, M. H. Ecogeographical differentiation patterns in Adonis sect. Consiligo (Ranunculaceae). Plant Syst. Evol. 211, 43–56 (1998).
Google Scholar 
Willner, W. et al. A higher-level classification of the Pannonian and western Pontic steppe grasslands (Central and Eastern Europe). Appl. Veg. Sci. 20, 143–158 (2017).PubMed 

Google Scholar 
Lange, D. Conservation and Sustainable Use of Adonis vernalis, a Medicinal Plant in International Trade (Landwirtschaftsverlag, Münster, 2000).Denisow, B., Wrzesień, M. & Cwener, A. Pollination and floral biology of Adonis vernalis L. (Ranunculaceae)—A case study of threatened species. Acta Soc. Bot. Pol. 83, 29–37 (2014).
Google Scholar 
Mitrenina, E. Y. et al. Karyotype and genome size in Adonis vernalis and Adonis volgensis. Turczaninowia 25, 5–15 (2022).
Google Scholar 
Zhai, W. et al. Chloroplast genomic data provide new and robust insights into the phylogeny and evolution of the Ranunculaceae. Mol. Phylogenet. Evol. 135, 12–21 (2019).CAS 
PubMed 

Google Scholar 
Bobrov, E. G. Genus 540. ADONIS L. In Flora Of The U.S.S.R., Volume VII, Ranales And Rhoeadales (ed. Komarov, V. L.) pp. 403–411 (Academy of Sciences of the U.S.S.R., 1970).Pisareva, V. V. et al. Changes in the landscape and climate of Eastern Europe in the Early Pleistocene. Stratigr. Geol. Correl. 27, 475–497 (2019).ADS 

Google Scholar 
Vislobokova, I., Tesakov, A. Early And Middle Pleistocene Of Northern Eurasia. In Encyclopedia Of Quaternary Science, Vol 4, 2nd edn (ed. Elias, S. A.) pp. 605–614 (Elsevier, Amsterdam, 2013).Hirsch, H. et al. High genetic diversity declines towards the geographic range periphery of Adonis vernalis, a Eurasian dry grassland plant. Plant Biol. 17, 1233–1241 (2015).CAS 
PubMed 

Google Scholar 
Kirschner, P. et al. Long-term isolation of European steppe outposts boosts the biome’s conservation value. Nat. Commun. 11, 1968 (2020).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Willner, W. et al. Long-term continuity of steppe grasslands in eastern Central Europe: Evidence from species distribution patterns and chloroplast haplotypes. J. Biogeogr. 48, 3104–3117 (2021).
Google Scholar 
Molnár, Z., Biró, M., Bartha, S. & Fekete, G. Past trends, present state and future prospects of Hungarian forest-steppes. In Eurasian Steppes. Ecological Problems And Livelihoods In A Changing World (eds Werger, M. J. A. & van Staalduinen, M. A.) pp. 209–252 (Springer, Dordrecht, 2012).
Google Scholar 
Liedtke, H. Die Nordischen Vereisungen in Mitteleuropa, 2nd edn. (Zentralausschuß für deutsche Landeskunde, Trier, 1981).Sizikova, A. O. & Zykina, V. S. The dynamics of the Late Pleistocene loess formation, Lozhok section, Ob loess Plateau, SW Siberia. Quat. Int. 365, 4–14 (2015).
Google Scholar 
Zykina, V. S. & Zykin, V. S. Pleistocene warming stages in Southern West Siberia: Soils, environment, and climate evolution. Quat. Int. 106–107, 233–243 (2003).
Google Scholar 
Shumilovskikh, L., Sannikov, P., Efimik, E., Shestakov, I. & Mingalev, V. V. Long-term ecology and conservation of the Kungur forest-steppe (pre-Urals, Russia): Case study Spasskaya Gora. Biodivers. Conserv. 30, 4061–4087 (2021).
Google Scholar 
Markova, A. & Puzachenko, A. Vertebrate records/Late Pleistocene of Northern Asia. In Encyclopedia of Quaternary Science Vol. 4 (ed. Elias, S. A.) 3158–3175 (Elsevier, Amsterdam, 2007).Gómez, C. & Espadaler, X. An update of the world survey of myrmecochorous dispersal distances. Ecography 36, 1193–1201 (2013).
Google Scholar 
Albert, A. et al. Seed dispersal by ungulates as an ecological filter: A trait-based meta-analysis. Oikos 124, 1109–1120 (2015).
Google Scholar 
Albert, A., Mårell, A., Picard, M. & Baltzinger, C. Using basic plant traits to predict ungulate seed dispersal potential. Ecography 38, 440–449 (2015).
Google Scholar 
Popescu, S.-M. et al. Late Quaternary vegetation and climate of SE Europe–NW Asia according to pollen records in three offshore cores from the Black and Marmara seas. Paleobiodivers. Paleoenviron. 101, 197–212 (2021).
Google Scholar 
Markova, A. K., Simakova, A. N. & Puzachenko, A. Y. Ecosystems of Eastern Europe at the time of maximum cooling of the Valdai glaciation (24–18 kyr BP) inferred from data on plant communities and mammal assemblages. Quat. Int. 201, 53–59 (2009).
Google Scholar 
Kajtoch, Ł et al. Phylogeographic patterns of steppe species in Eastern Central Europe: A review and the implications for conservation. Biodivers. Conserv. 25, 2309–2339 (2016).
Google Scholar 
Kropf, M., Bardy, K., Höhn, M. & Plenk, K. Phylogeographical structure and genetic diversity of Adonis vernalis L. (Ranunculaceae) across and beyond the Pannonian region. Flora 262, 151497 (2020).
Google Scholar 
Plenk, K., Bardy, K., Höhn, M., Thiv, M. & Kropf, M. No obvious genetic erosion, but evident relict status at the westernmost range edge of the Pontic-Pannonian steppe plant Linum flavum L. (Linaceae) in Central Europe. Ecol. Evol. 7, 6527–6539 (2017).PubMed 
PubMed Central 

Google Scholar 
Magyari, E. K. et al. Late Pleniglacial vegetation in eastern-central Europe: Are there modern analogues in Siberia? Quat. Sci. Rev. 95, 60–79 (2014).ADS 

Google Scholar 
Doležel, J., Greilhuber, J. & Suda, J. Estimation of nuclear DNA content in plants using flow cytometry. Nat. Protoc. 2, 2233–2244 (2007).PubMed 

Google Scholar 
Shaw, J., Lickey, E. B., Schilling, E. E. & Small, R. L. Comparison of whole chloroplast genome sequences to choose noncoding regions for phylogenetic studies in angiosperms: The tortoise and the hare III. Am. J. Bot. 94, 275–288 (2007).CAS 
PubMed 

Google Scholar 
Shaw, J. et al. The tortoise and the hare II: Relative utility of 21 noncoding chloroplast DNA sequences for phylogenetic analysis. Am. J. Bot. 92, 142–166 (2005).CAS 
PubMed 

Google Scholar 
Heckenhauer, J., Barfuss, M. H. J. & Samuel, R. Universal multiplexable matK primers for DNA barcoding of angiosperms. Appl. Plant Sci. 4, 1500137 (2016).
Google Scholar 
White, T. J., Bruns, T., Lee, S. & Taylor, J. Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics. In PCR Protocols: A Guide to Methods and Applications (eds Innis, M. A. et al.) pp. 315–322 (Academic Press, Cambridge, 1990).
Google Scholar 
Simmons, M. P. & Ochoterena, H. Gaps as characters in sequence-based phylogenetic analyses. Syst. Biol. 49, 369–381 (2000).CAS 
PubMed 

Google Scholar 
Müller, K. SeqState: Primer design and sequence statistics for phylogenetic DNA datasets. Appl. Bioinformat. 4, 65–69 (2005).
Google Scholar 
Swofford, D. L. PAUP*. Software. https://paup.phylosolutions.com/.Kozlov, A., Darriba, D., Flouri, T., Morel, B. & A.,. Stamatakis RAxML-NG: A fast, scalable, and user-friendly tool for maximum likelihood phylogenetic inference. Bioinformatics 35, 4453–4455 (2019).CAS 
PubMed 
PubMed Central 

Google Scholar 
Bouckaert, R. et al. BEAST 2.5: An advanced software platform for Bayesian evolutionary analysis. PLoS Comput. Biol. 15, e1006650 (2019).CAS 
PubMed 
PubMed Central 

Google Scholar 
Heled, J. & Drummond, A. J. Calibrated tree priors for relaxed phylogenetics and divergence time estimation. Syst. Biol. 61, 138–149 (2012).PubMed 

Google Scholar 
Drummond, A. J., Ho, S. Y. W., Phillips, M. J. & Rambaut, A. Relaxed phylogenetics and dating with confidence. PLoS Biol. 4, e88 (2006).PubMed 
PubMed Central 

Google Scholar 
Rambaut, A., Drummond, A. J., Xie, D., Baele, G. & Suchard, M. A. Posterior summarization in Bayesian phylogenetics using Tracer 1.7. Syst. Biol. 67, 901–904 (2018).CAS 
PubMed 
PubMed Central 

Google Scholar 
Rambaut, A. FigTree. Software. http://tree.bio.ed.ac.uk/software/figtree/.Wendler, N. et al. Unlocking the secondary gene-pool of barley with next-generation sequencing. Plant Biotechnol. J. 12, 1122–1131 (2014).ADS 
CAS 
PubMed 

Google Scholar 
Eaton, D. A. R. & Overcast, I. ipyrad: Interactive assembly and analysis of RADseq datasets. Bioinformatics 36, 2592–2594 (2020).CAS 
PubMed 

Google Scholar 
Weiß, C. L., Pais, M., Cano, L. M., Kamoun, S. & Burbano, H. A. nQuire: A statistical framework for ploidy estimation using next generation sequencing. BMC Bioinform. 19, 122 (2018).
Google Scholar 
Corrêa dos Santos, R. A., Goldman, G. H. & Riaño-Pachón, D. M. ploidyNGS: Visually exploring ploidy with next generation sequencing data. Bioinformatics 33, 2575–2576 (2017).
Google Scholar 
Frichot, E. & François, O. LEA: An R package for landscape and ecological association studies. Methods Ecol. Evol. 6, 925–929 (2015).
Google Scholar 
R Core Team. R. A language and environment for statistical computing. Software. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/.Gruber, B., Unmack, P. J., Berry, O. F. & Georges, A. dartR: An R package to facilitate analysis of SNP data generated from reduced representation genome sequencing. Mol. Ecol. Resour. 18, 691–699 (2018).PubMed 

Google Scholar 
Danecek, P. et al. The variant call format and VCFtools. Bioinformatics 27, 2156–2158 (2011).CAS 
PubMed 
PubMed Central 

Google Scholar 
Minh, B. Q. et al. IQ-TREE 2: New models and efficient methods for phylogenetic inference in the genomic era. Mol. Biol. Evol. 37, 1530–1534 (2020).CAS 
PubMed 
PubMed Central 

Google Scholar 
Guindon, S. et al. New algorithms and methods to estimate maximum-likelihood phylogenies: Assessing the performance of PhyML 3.0. Syst. Biol. 59, 307–321 (2010).CAS 
PubMed 

Google Scholar 
Minh, B. Q., Nguyen, M. A. T. & von Haeseler, A. Ultrafast approximation for phylogenetic bootstrap. Mol. Biol. Evol. 30, 1188–1195 (2013).CAS 
PubMed 
PubMed Central 

Google Scholar 
Huson, D. H. & Bryant, D. Application of phylogenetic networks in evolutionary studies. Mol. Biol. Evol. 23, 254–267 (2005).PubMed 

Google Scholar 
Kalinowski, S. T. HP-RARE 1.0: A computer program for performing rarefaction on measures of allelic richness. Mol. Ecol. Notes 5, 187–189 (2005).CAS 

Google Scholar  More

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

Source: World Bank

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

Google Scholar  More

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