Philippa Kaur

Kollas, C., Körner, C. & Randin, C. F. Spring frost and growing season length co-control the cold range limits of broad-leaved trees. J. Biogeogr. 41, 773–783 (2014).Article 

Google Scholar 
Lenz, A., Hoch, G., Körner, C. & Vitasse, Y. Convergence of leaf-out towards minimum risk of freezing damage in temperate trees. Funct. Ecol. 30, 1480–1490 (2016).Article 

Google Scholar 
Körner, C. et al. Where, why and how? Explaining the low-temperature range limits of temperate tree species. J. Ecol. 104, 1076–1088 (2016).Article 
CAS 

Google Scholar 
Körner, C. Plant adaptation to cold climates. F1000Research 5, 2769 (2016).Article 

Google Scholar 
Vitra, A., Lenz, A. & Vitasse, Y. Frost hardening and dehardening potential intemperate trees from winter to budburst. New Phytol. 216, 113–123 (2017).PubMed 
Article 

Google Scholar 
Dy, G. & Payette, S. Frost hollows of the boreal forest as extreme environments for black spruce tree growth. Can. J. For. Res. 37, 492–504 (2007).Article 

Google Scholar 
Marquis, B., Bergeron, Y., Simard, M. & Tremblay, F. Growing-season frost is a better predictor of tree growth than mean annual temperature in boreal mixedwood forest plantations. Glob. Change Biol. 26, 6537–6554 (2020).ADS 
Article 

Google Scholar 
Hufkens, K. et al. Ecological impacts of a widespread frost event following early spring leaf-out. Glob. Change Biol. 18, 2365–2377 (2012).ADS 
Article 

Google Scholar 
Guo, X., Khare, S., Silvestro, R. & Rossi, S. Minimum spring temperatures at the provenance origin drive leaf phenology in sugar maple populations. Tree Physiol. 40, 1639–1647 (2020).CAS 
PubMed 
Article 

Google Scholar 
Marquis, B., Bergeron, Y., Simard, M. & Tremblay, F. Probability of spring frosts, not growing degree-days, drives onset of spruce bud burst in plantations at the boreal-temperate forest ecotone. Front. Plant Sci. 11, 1031 (2020).PubMed 
PubMed Central 
Article 

Google Scholar 
Mahony, C. R., Cannon, A. J., Wang, T. & Aitken, S. N. A closer look at novel climates: New methods and insights at continental to landscape scales. Glob. Change Biol. 23, 3934–3955 (2017).ADS 
Article 

Google Scholar 
Roman-Palacios, C. & Wiens, J. Recent responses to climate change reveal the drivers of species extinction and survival. Proc. Natl. Acad. Sci. U.S.A. 117, 4211–4217 (2020).CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Chuine, I. A unified model for budburst of trees. J. Theor. Biol. 207, 337–347 (2000).ADS 
CAS 
PubMed 
Article 

Google Scholar 
Horvath, P. et al. Improving the representation of high-latitude vegetation distribution in dynamic global vegetation models. Biogeosciences 18, 95–112 (2021).ADS 
CAS 
Article 

Google Scholar 
Wullschleger, S. D. et al. Plant functional types in Earth system models: Past experiences and future directions for application of dynamic vegetation models in high-latitude ecosystems. Ann. Bot. Lond. 114, 1–16 (2014).CAS 
Article 

Google Scholar 
Deser, C. et al. Insights from earth system model initial-condition large ensembles and future prospects. Nat. Clim. Change 10, 277–286 (2020).ADS 
Article 

Google Scholar 
Machete, R. L. & Smith, L. A. Demonstrating the value of larger ensembles in forecasting physical systems. Tellus A Dyn. Meteorol. Oceanogr. 68, 28393 (2016).Article 

Google Scholar 
Deser, C., Phillips, A., Bourdette, V. & Teng, H. Uncertainty in climate change projections: The role of internal variability. Clim. Dyn. 38, 527–546 (2012).Article 

Google Scholar 
Leduc, M. et al. The ClimEx project: A 50-member ensemble of climate change projections at 12-km resolution over Europe and Northeastern North America with the Canadian regional climate model (CRM5). J. Appl. Meteor. Climatol. 58, 663–693 (2019).ADS 
Article 

Google Scholar 
Innocenti, S., Mailhot, A., Leduc, M., Cannon, A. J. & Frigon, A. Projected changes in the probability distributions, seasonality, and spatiotemporal scaling of daily and subdaily extreme precipitation simulated by a 50-member ensemble over Northeastern North America. J. Geophys. Res. Atmos. 124, 19 (2019).Article 

Google Scholar 
Kay, J. E. et al. The community earth system model (CESM) large ensemble project: A community resource for studying climate change in the presence of internal climate variability. Bull. Am. Meteor. Soc. 96, 1333–1349 (2015).ADS 
Article 

Google Scholar 
Thompson, D. W. J., Barnes, E. A., Deser, C., Foust, W. E. & Phillips, A. S. Quantifying the role of internal climate variability in future climate trends. J. Clim. 28, 6443–6456 (2015).ADS 
Article 

Google Scholar 
Kumar, D. & Ganguly, A. R. Intercomparison of model response and internal variability across climate model ensembles. Clim. Dyn. 51, 207–219 (2018).Article 

Google Scholar 
Gu, L. et al. The contribution of internal climate variability to climate change impacts on droughts. Sci. Total Environ. 684, 229–246 (2019).ADS 
CAS 
PubMed 
Article 

Google Scholar 
Mittermeier, M., Braun, M., Hofstätter, M., Wang, Y. & Ludwig, R. Detecting climate change effects on Vb cyclones in a 50-member single-model ensemble using machine learning. Geophys. Res. Lett. 46, 14653–14661 (2019).ADS 
Article 

Google Scholar 
Fyfe, J. C. et al. Large near-term projected snowpack loss over the western United States. Nat. Commun. 8, 14996 (2017).ADS 
CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Wickham, H. ggplot2: Elegant Graphics for Data Analysis (Springer, 2016).MATH 
Book 

Google Scholar 
Liu, Q. et al. Extension of the growing season increases vegetation exposure to frost. Nat. Commun. 9, 426 (2018).ADS 
PubMed 
PubMed Central 
Article 
CAS 

Google Scholar 
Zohner, C. M. et al. Late-spring frost risk between 1959 and 2017 decreased in North America but increased in Europe and Asia. Proc. Natl. Acad. Sci. U.S.A. 117, 12192–12200 (2020).CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Ma, Q., Huang, J.-G., Hänninen, H. & Berninger, F. Divergent trends in the risk of spring frost damage to trees in Europe with recent warming. Glob. Change Biol. 25, 351–360 (2018).ADS 
Article 

Google Scholar 
Cannell, M. G. R. & Smith, R. I. Climatic warming, spring budburst and forest damage on trees. J. Appl. Ecol. 23, 177–191 (1986).Article 

Google Scholar 
Morin, X. & Chuine, I. Will tree species experience increased frost damage due to climate change because of changes in leaf phenology? Can. J. For. Res. 44, 1555–1565 (2014).Article 

Google Scholar 
Fu, Y. et al. Progress in plant phenology modeling under global climate change. Sci. China Earth Sci. 63, 1237 (2020).ADS 
Article 

Google Scholar 
Gill, A. L. et al. Changes in autumn senescence in northern hemisphere deciduous trees: A meta-analysis of autumn phenology studies. Ann Bot. 116, 875–888 (2015).CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Perrin, M., Rossi, S. & Isabel, S. N. Synchronisms between bud and cambium phenology in black spruce: Early-flushing provenances exhibit early xylem formation. Tree Physiol. 37, 593–603 (2017).PubMed 
Article 

Google Scholar 
Guo, X. et al. Common-garden experiment reveals clinical trends of bud phenology in black spruce populations from a latitudinal gradient in the boreal forest. J. Ecol. https://doi.org/10.1111/1365-2745.13582 (2021).Article 

Google Scholar 
Usmani, A. et al. Ecotypic differentiation of black spruce populations: Temperature triggers bud burst but not bud set. Trees 34, 1313–1321 (2020).Article 

Google Scholar 
Buttò, V., Rozenberg, P., Deslauriers, A., Rossi, S. & Morin, H. Environmental and developmental factors driving xylem anatomy and micro-density in black spruce. New Phytol. 230, 957. https://doi.org/10.1111/nph.17223 (2021).CAS 
Article 
PubMed 

Google Scholar 
Keenan, T. F. & Richardson, A. D. The timing of autumn senescence is affected by the timing of spring phenology: Implications for predictive models. Glob. Change Biol. 21, 2634–2641 (2015).ADS 
Article 

Google Scholar 
Zani, D., Crowther, T. W., Mo, L., Renner, S. S. & Zohner, C. M. Increased growing-season productivity drives earlier autumn leaf senescence in temperate trees. Science 370, 1066–1071 (2020).ADS 
CAS 
PubMed 
Article 

Google Scholar 
Silvestro, R. et al. From phenology to forest management: Ecotypes selection can avoid early or late frosts, but not both. For. Ecol. Manage. 436, 21–26 (2019).Article 

Google Scholar 
D’Orangeville, L. et al. Beneficial effects of climate warming on boreal tree growth may be transitory. Nat. Commun. 9, 3213 (2018).ADS 
PubMed 
PubMed Central 
Article 
CAS 

Google Scholar 
Menezes-Silva, P. E. et al. Different ways to die in a changing world: Consequences of climate change for tree species performance and survival through an ecophysiological perspective. Ecol. Evol. 9, 11979–11999 (2019).PubMed 
PubMed Central 
Article 

Google Scholar 
Schwarz, J. A., Saha, S. & Bauhus, J. Potential of forest thinning to mitigate drought stress: A meta-analysis. For. Ecol. Manage. 380, 261–273 (2016).Article 

Google Scholar 
Marquis, B., Bergeron, Y., Simard, M. & Tremblay, F. Disentangling the effect of topography and microtopography on near-ground growing-season frosts at the boreal-temperate forest ecotone (Québec, Canada). New For. https://doi.org/10.1007/s11056-021-09840-7 (2021).Article 

Google Scholar 
Marquis, B. growth stagnation of planted spruce in boreal mixedwoods: Importance of landscape, microsite, and growing-season frosts. For. Ecol. Manage. 479, 118533 (2021).Article 

Google Scholar 
Pedlar, J. et al. The implementation of assisted migration in Canadian forests. For. Chron. 87, 766–777 (2011).Article 

Google Scholar 
Marris, E. Planting the forest for the future. Nature 459, 906–908 (2009).CAS 
PubMed 
Article 

Google Scholar 
Klisz, M. et al. Limitations at the limit? Diminishing of genetic effects in Norway spruce provenance trials. Front. Plant Sci. 10, 306 (2019).PubMed 
PubMed Central 
Article 

Google Scholar 
Pedlar, J. H., McKenney, D. W. & Lu, P. Critical seed transfer distances for selected tree species in eastern North America. J. Ecol. https://doi.org/10.1111/1365-2745.13605 (2021).Article 

Google Scholar 
Prudhomme, G. O. et al. Ecophysiology and growth of white spruce seedlings from various seed sources along a climatic gradient support the need for assisted migration. Front. Plant Sci. 8, 2214 (2017).Article 

Google Scholar 
Girardin, M. P. et al. Annual aboveground carbon uptake enhancements from assisted gene flow in boreal black spruce forests are not long-lasting. Nat. Commun. 12, 1169 (2021).ADS 
CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Zohner, C. M., Mo, L., Sebald, V. & Renner, S. S. Leaf-out in northern ecotypes of wide-ranging trees requires less spring warming, enhancing the risk of spring frost damage at cold range limits. Glob. Ecol. Biogeogr. 29, 1065–1072 (2020).Article 

Google Scholar 
Saucier, J.-P., Baldwin, K., Krestov, P. & Jorgenson, T. Boreal forests. In Routledge Handbook of Forest Ecology (eds Peh, K.S.-H. et al.) 7–29 (Routledge, 2015).
Google Scholar 
Hutchinson, M. F. et al. Development and testing of Canada-wide interpolated spatial models of daily minimum-maximum temperature and precipitation for 1961–2003. J. Appl. Meteorol. Climatol. 48, 725–741 (2009).ADS 
Article 

Google Scholar 
Gennaretti, F., Sangelantoni, L. & Grenier, P. Toward daily climate scenarios for Canadian Arctic coastal zones with more realistic temperature-precipitation interdependence. J. Geophys. Res. Atmos. 120, 862–877 (2015).Article 

Google Scholar 
Mpelasoka, F. S. & Chiew, F. H. S. Influence of rainfall scenario construction methods on runoff projections. J. Hydrometeorol. 10, 1168–1183 (2009).ADS 
Article 

Google Scholar 
Giorgi, F. Thirty years of regional climate modeling: Where are we and where are we going next? J. Geophys. Res. 124, 5696–5723 (2019).Article 

Google Scholar 
Laughlin, G. P. & Kalma, J. D. Frost hazard assessment from local weather and terrain data. Agric. For. Meteorol. 40, 1–16 (1987).ADS 
Article 

Google Scholar 
Sørland, S. L., Schar, C., Luthi, D. & Kjellstrom, E. Bias patterns and climate change signals in GCM-RCM model chains. Environ. Res. Lett. 13, 074017 (2018).ADS 
Article 

Google Scholar 
von Trentini, F., Aalbers, E. E., Fischer, E. M. & Ludwig, R. Comparing interannual variability in three regional single-model initial-condition large ensembles (smiles) over Europe. Earth Syst. Dyn. 11, 1013–1031 (2020).ADS 
Article 

Google Scholar 
Hänninen, H. & Pelkonen, H. Effects of temperature on dormancy release in Norway spruce and Scots pine seedlings. Silva Fenn. 22, 5357 (1988).
Google Scholar 
Lechowicz, M. J. Why do temperate deciduous trees leaf out at different times? Adaptation and ecology of forest communities. Am. Nat. 124, 821–842 (1984).Article 

Google Scholar 
Olson, M. S. et al. The adaptive potential of Populus balsamifera L. to phenology requirements in a warmer global climate. Mol. Ecol. 22, 1214–1230 (2013).CAS 
PubMed 
Article 

Google Scholar 
Hawkins, C. D. B. & Dhar, A. Spring bud phenology of 18 Betula papyrifera populations in British Columbia. Scand. J. For. Res. 27, 507–519 (2012).Article 

Google Scholar 
Bronson, D. R., Gower, S. T., Tanner, M. & Van Herk, I. Effect of ecosystem warming on boreal black spruce bud burst and shoot growth. Glob. Change Biol. 15, 1534–1543 (2009).ADS 
Article 

Google Scholar 
Basler, D. Evaluating phenological models for the prediction of leaf-out dates in six temperate tree species across central Europe. Agric. For. Meteorol. 217, 10–21 (2016).ADS 
Article 

Google Scholar 
Man, R., Lu, P. & Dang, Q.-L. Insufficient chilling effects vary among boreal tree species and chilling duration. Front. Plant Sci. https://doi.org/10.3389/fpls.2017.01354 (2017).Article 
PubMed 
PubMed Central 

Google Scholar 
R Development Core Team. R: A Language and Environment for Statistical Computing. (R Foundation for Statistical Computing, 2020). http://www.R-project.org. More

Farmers at a Great Green Wall site in Niger. Researchers have found that the project is not always benefiting the most vulnerable people.Credit: Boureima Hama/AFP/Getty

It’s now 15 years since the African Union gave its blessing to Africa’s Great Green Wall, one of the world’s most ambitious ecological-restoration schemes. The project is intended to combat desertification across the width of Africa, and spans some 8,000 kilometres, from Senegal to Djibouti. Its ambition is staggering: it aims to restore 100 million hectares of degraded land by 2030, capturing 250 million tonnes of carbon dioxide and creating 10 million jobs in the process. But it continues to struggle.An assessment two years ago by independent experts commissioned by the United Nations stated that somewhere between 4% and 20% of the restoration target had been achieved (go.nature.com/39zqgkr). That figure has not changed, according to the latest edition of Global Land Outlook (go.nature.com/3kdjtw5) from the UN Convention to Combat Desertification (UNCCD), out last week. Equally concerning is the fact that funding for the project continues to lag. Africa’s governments and international donors need to find around US$30 billion to reach the 100-million-hectare target. So far, $19 billion has been raised.A pandemic — and now a cost-of-living crisis — has placed demands on all governments, and that means countries might be expected to reduce their green-wall commitments. But the project continues to be weighed down by other difficulties, including the complex system through which it is funded and governed, as well as how its success is measured. These problems can and must be fixed, otherwise it will struggle to achieve its goals.One potential solution — improved metrics — comes from an analysis published last year by Matthew Turner at the University of Wisconsin–Madison and his colleagues (M. D. Turner et al. Land Use Policy 111, 105750; 2021). The researchers explored limitations in the Great Green Wall project metrics by assessing the impact of World Bank funding from 2006 to 2020. As their work indicates, definitions of success depend on which measure is used.In Niger, for example, green-wall projects could be said to be succeeding if measured by the area of eroded soil that has been recovered or by the number of trees that have been planted. But the authors report that these gains were not necessarily benefiting the most vulnerable people. In places, women were being excluded from employment in green-wall projects, and in some cases, local administrations looked to privatize restored land that might instead have been owned by everyone in a community.Broader problems with metrics are highlighted in the UN’s latest land-degradation report. This estimates that nearly half of the land that has been pledged for restoration worldwide will be planted predominantly with fast-growing trees and plants. This will provide only a fraction of the ecosystem services produced by forests that are allowed to naturally regenerate, including significantly less carbon storage, groundwater recharge and wildlife habitat.The Great Green Wall project also needs more predictable funding and more transparent governance. The project was conceived by Africa’s leaders for the benefit of the continent’s people, on the basis of warnings from scientists about the risks of desertification and land degradation. The original idea was not brought to Africa by international donors, as is often the case in international science-based development projects. But it still relies on donor financing, and lots of it — and that brings other problems, among them coordination challenges.The project is the responsibility of an organization set up by the African Union called the Pan African Agency of the Great Green Wall, based in Nouakchott, Mauritania. But some donors, such as the European Union and the World Bank, are not providing most of their Great Green Wall funding through this agency. Instead, they often deal directly with individual governments, because this gives them more control over how their money is spent. It is unfair to expect the Pan African Agency to coordinate a raft of donors doing one-on-one deals with individual countries. Bypassing the Pan African Agency also creates a problem for transparency, because it makes it harder for the African Union to determine precisely who is funding what.In January 2021, at an international biodiversity summit hosted by France, Emmanuel Macron, the French president, announced that the Great Green Wall would receive an extra $14 billion in funding for 5 years. He also said that a new body, called the Great Green Wall Accelerator, based in Bonn, Germany, would be responsible for pulling together funding pledges and tracking progress against targets. This is well-intentioned, but the accelerator needs to coordinate its work with the Pan African Agency. It is not yet clear how this will happen.A potentially more transformative solution was proposed two years ago by a group of UN-appointed experts. They recommended that a single trust fund be set up that all donors could contribute to and through which they could decide funding priorities together. Regrettably, this has not happened, and observers say it is not likely to happen in the current climate.This month, the international community will come together in Abidjan, Côte d’Ivoire, for the 15th conference of the parties to the UNCCD. The green wall’s funders and participating countries will all be there. If a single trust fund is off the table, they must work together to find a better way to coordinate their green-wall project activities. It is also essential that they study the findings of Turner and colleagues’ review. Along with a focus on existing metrics, the Great Green Wall needs evaluation criteria that take better account of the needs of all people in participating countries, particularly the most vulnerable. More

Dordevic, B. & Dasic, T. Water storage reservoirs and their role in the development, utilization and protection of catchment. SPATIUM Int. Rev. 24, 9–15 (2011).
Google Scholar 
European Community. Directive 2000/60/EC of the European Parliament and of the Council of 23 October 2000 establishing a framework for Community action in the field of water policy. Official J. Eur. Union. 5, L327 (2000).

Google Scholar 
Riley, W. D. et al. Small water bodies in Great Britain and Ireland: Ecosystem function, human-generated degradation, and options for restorative action. Sci. Total Environ. 645, 1598–1616. https://doi.org/10.1016/j.scitotenv.2018.07.243 (2018).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Kujawa, K., Arczyńska-Chudy, E., Janku, K. & Mana, M. Effect of buffer zone structure on diversity of aquatic vegetation in farmland water bodies. Pol. J. Ecol. 68(4), 263–282. https://doi.org/10.3161/15052249PJE2020.68.4.001 (2021).Article 

Google Scholar 
Akasaka, M., Takamura, N., Mitsuhashi, H. & Kadono, Y. Effects of land use on aquatic macrophyte diversity and water quality of ponds. Fresh Biol. 55, 909–922 (2010).CAS 
Article 

Google Scholar 
Pukacz, A., Pełechaty, M., Pełechata, A. & Siepak, M. The differential cover of submerged vegetation vs habitat conditions in the lakes of the Lubuskie Region. Limnol. Rev. 7(2), 95–100 (2007).
Google Scholar 
Scheffer, M., Hosper, S. H., Meijer, M. L., Moss, B. & Jeppesen, E. Alternative equilibria in shallow lakes. Trends Ecol. Evol. 8, 275–279 (1993).CAS 
Article 

Google Scholar 
Scheffer, M. & Jeppesen, E. Alternative stable states. Ecol. Stud. 131, 397–405 (1998).Article 

Google Scholar 
Beck, M. W., Tomcko, C., Valley, R. D. & Staples, D. F. Analysis of macrophyte indicator variation as a function of sampling, temporal and stressor effects. Ecol. Indic. 46, 322–335 (2014).Article 

Google Scholar 
Celewicz-Gołdyn, S. & Kuczyńska-Kippen, N. Ecological value of macrophyte cover in creating habitat for microalgae (diatoms) and zooplankton (rotifers and crustaceans) in small field and forest water bodies. PLoS ONE 12(5), e0177317. https://doi.org/10.1371/journal.pone.0177317 (2017).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Wilk-Woźniak, E. et al. Effects of the environs of waterbodies on aquatic plants in oxbow lakes (habitat 3150). Ecol. Ind. 98, 736–742. https://doi.org/10.1016/j.ecolind.2018.11.025 (2019).CAS 
Article 

Google Scholar 
Bedla, D. & Misztal, A. Changeability of chemistry of small water reservoirs with diversified use structure of the adjoining areas. Annu. Set Environ. Protect. 16, 421–439 (2014) (ISSSN 1506-218X (in Polish)).
Google Scholar 
Mozgawa, J. Photointerpretation analysis of landscape structure in lake watersheds of Suwałki Landscape. Park. Ekol. Pol. 41, 53–74 (1993).
Google Scholar 
Mioduszewski, W. Small water reservoirs on agricultural areas. Wieś Jutra Zakład Zasobów Wodnych Instytut Melioracji i Użytków Zielonych Falenty. 10(123), 32–34 (2008) (in Polish).
Google Scholar 
Gołdyn, B., Kowalczewska-Madura, K. & Celewicz-Gołdyn, S. Drought and deluge: Influence of environmental factors on water quality of kettle holes in two subsequent years with different precipitation. Limnologica 54, 14–22. https://doi.org/10.1016/j.limno.2015.07.002 (2015).CAS 
Article 

Google Scholar 
Bylak, A. et al. Small stream catchment in a developing city context: The importance of land cover changes on the ecological status of streams and the possibilities for providing ecosystem services. Sci. Total Environ. 815, 151974. https://doi.org/10.1016/j.scitotenv.2021.151974 (2022).ADS 
CAS 
Article 
PubMed 

Google Scholar 
Goyal, V. C. et al. Ecological health and water quality of village ponds in the subtropics limiting their use for water supply and groundwater recharge. J. Environ. Manage. 277, 111450. https://doi.org/10.1016/j.jenvman.2020.111450 (2021).CAS 
Article 
PubMed 

Google Scholar 
Kuczyńska-Kippen, N., Spoljar, M., Mleczek, M. & Zhang, Ch. Elodeides, but not helophytes, increase community diversity and reduce trophic state: Case study with rotifer indices in field ponds. Ecol. Ind. 128, 107829. https://doi.org/10.1016/j.ecolind.2021.107829 (2021).Article 

Google Scholar 
Novikmec, M. et al. Ponds and their catchments: size relationships and influence of land use across multiple spatial scales. Hydrobiologia 774, 155–166. https://doi.org/10.1007/s10750-015-2514-8 (2016).Article 

Google Scholar 
Dudzińska, A., Szpakowska, B. & Pajchrowska, M. Influence of land development on the ecological status of small water bodies. Oceanol. Hydrobiol. Stud. 49(4), 345–353. https://doi.org/10.1515/ohs-2020-0030 (2020).Article 

Google Scholar 
Szpakowska, B. Occurrence and Role of Organic Compounds Dissolved in Surface and Ground Waters of Agricultural Landscape (Wydawnictwo Uniwersytetu Mikołaja Kopernika, 1999) (in Polish).
Google Scholar 
Wysocka-Czubaszek, A. & Banaszuk, P. Migration of nitrogen compounds and the riparian zones in the Upper Narew Valley. Acta Agroph. 2(1), 349–354 (2003) (in Polish).
Google Scholar 
Szpakowska, B., Świerk, D., Pajchrowska, M. & Gołdyn, R. Verifying the usefulness of macrophytes as an indicator of the status of small waterbodies. Sci. Total Environ. 798, 149279. https://doi.org/10.1016/j.scitotenv.2021.149279 (2021).ADS 
CAS 
Article 
PubMed 

Google Scholar 
Hermanowicz, W., Dojlido, J., Dożańska, W., Koziorowski, B. & Zerbe, J. Physico-Chemical Examinations of Water and Sludge (Arkady, 1999) (in Polish).
Google Scholar 
Aynur, M., Liming, N. & Fang, Y. Spatial evaluation of environmental suitability for human settlement of Kashgar, Northwest China. Fresenius Environ. Bull. 27(9), 5899–5907 (2018).
Google Scholar 
Pham, L., Brabyn, L. & Ashrof, S. Combining Quick Bird, LiDAR and GIS topography indices to identify a single native tree species in a complex landscape using an object-based classification approach. Int. J. Appl. Earth Obs. Geoinf. 50, 187–197. https://doi.org/10.1016/j.jag.2016.03.015 (2016).ADS 
Article 

Google Scholar 
Lepš, J. & Šmilauer, P. Multivariate Analysis of Ecological Data Using CANOCO (Cambridge Univeristy Press, 2003) (ISBN 9780521891080).Book 

Google Scholar 
Mohamed, Z. A. Macrophytes-cyanobacteria allelopathic interactions and their implications for water resources management—A review. Limnologica 63, 122–132. https://doi.org/10.1016/j.limno.2017.02.006 (2017).CAS 
Article 

Google Scholar 
Koc, J. Problems of small reservoirs protection on rural areas. In Problemy ochrony i użytkowania obszarów wiejskich o dużych walorach przyrodniczych (eds Radwan, S. & Lorkiewicz, Z.) 151–156 (UMCS, 2000).
Google Scholar 
Bell, V. A. et al. Long term simulations of macronutrients (C, N and P) in UK freshwaters. Sci. Total Environ. 776, 145813. https://doi.org/10.1016/j.scitotenv.2021.145813 (2021).ADS 
CAS 
Article 

Google Scholar 
Winton, R. S. et al. Anthropogenic influences on Zambian water quality: Hydropower and land-use change. Environ. Sci. Process. Impacts. 23, 981–994. https://doi.org/10.1039/D1EM00006C (2021).CAS 
Article 
PubMed 

Google Scholar 
Wang, Y. et al. Effect and risk assessment of animal manure pollution on Huaihe River Basin, China. Chin. Geogr. Sci. 31, 751–764. https://doi.org/10.1007/s11769-021-1222-8 (2021).Article 

Google Scholar 
Lawniczak-Malińska, A., Ptak, M., Celewicz, S. & Choiński, A. Impact of lake morphology and shallowing on the rate of overgrowth in hard-water eutrophic lakes. Water 10(12), 1827. https://doi.org/10.3390/w10121827 (2018).CAS 
Article 

Google Scholar 
Bosiacka, B., Pacewicz, K. & Pieńkowski, P. Spatial analysis of plant species distribution among small water bodies in an agricultural landscape. Acta Agrobot. 61(2), 93–101. https://doi.org/10.5586/aa.2008.037 (2008).Article 

Google Scholar 
Joniak, T., Kuczyńska-Kippen, N. & Gąbka, M. Effect of agricultural landscape characteristics on the hydrobiota structure in small water bodies. Hydrobiologia 793, 121–133. https://doi.org/10.1007/s10750-016-2913-5 (2017).CAS 
Article 

Google Scholar 
Barling, R. D. & Moore, I. D. Role of buffer strips in management of waterway pollution: A review. Environ. Manage. 18, 543–558. https://doi.org/10.1007/BF02400858 (1994).ADS 
Article 

Google Scholar 
Blanco-Cangui, H., & Lal, R. Buffer strips. In Principles of Soil Conservation and Management (eds Blanco, H. & Lal, R.) 223–257 (Springer, Netherlands, 2008).
Google Scholar 
Borin, M., Passoni, M., Thiene, M. & Tempesta, T. Multiple functions of buffer strips in farming areas. Eur. J. Agron. 32, 103–111. https://doi.org/10.1016/j.eja.2009.05.003 (2010).Article 

Google Scholar 
Ghobrial, M. G., Nassr, H. S. & Kamil, A. W. Bioactivity effect of two macrophyte extracts on growth performance of two bloom-forming cyanophytes. Egypt. J. Aquat. Res. 41(1), 69–81. https://doi.org/10.1016/j.ejar.2015.01.001 (2015).Article 

Google Scholar 
Lacas, J. G., Voltz, M., Gouy, V., Carluer, N. & Gril, J. J. Using grassed strips to limit pesticide transfer to surface water: A review. Agron. Sustain. Dev. 25, 253–266 (2005).CAS 
Article 

Google Scholar 
Świerk, D. & Szpakowska, B. An ecosystem valuation method for small water bodies. Ecol. Chem. Eng. S. 20(2), 397–418. https://doi.org/10.2478/eces-2013-0029 (2013).Article 

Google Scholar 
Marszałek, M., Kowalski, Z. & Makara, A. The possibility of contamination of water-soil environment as a results of the use of pig slurry. Ecol. Chem. Eng. S. 26(2), 313–330. https://doi.org/10.1515/eces-219-0022 (2019).Article 

Google Scholar 
Micek, G., Górecki, J. & Neo, H. Relations: Company and its local milieu in the context of foreign direct investment in Polish pig industry. In Człowiek i Środowisko (eds Górka, Z. & Zborowski, A.) 297–308 (UJ Kraków, 2009).
Google Scholar 
OECD. Agriculture Trade and The Environment: The Pig Sector (OECD, 2003).
Google Scholar 
Barałkiewicz, D. et al. Storm water contamination and its effect on the quality of urban surface waters. Environ. Monit. Assess. 186(10), 6789–6803. https://doi.org/10.1007/s10661-014-3889-0 (2014).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
So Fijanic, A., Hulley, M. & Loock, D. Stormwater quality assessment and management for the town of jasper in Alberta, Canada. Water Qual. Res. J. Can. 56(1), 45–56. https://doi.org/10.2166/wqrj.2021.012 (2021).CAS 
Article 

Google Scholar 
Zubala, T. Assessment of the variability of rainwater quality and the functioning of retention reservoirs in the urban area. Rocznik Ochrona Środowiska 22(2), 840–858 (2020).
Google Scholar 
Czerniawski, R., Sługocki, L., Krepski, T., Wilczak, A. & Pietrzak, K. Spatial changes in invertebrate structures as a factor of strong human activity in the bed and catchment area of a small urban stream. Water 12(3), 913. https://doi.org/10.3390/w12030913 (2020).CAS 
Article 

Google Scholar 
Gołdyn, R. et al. Influence of stormwater runoff on macroinvertebrates in a small urban river and a reservoir. Sci. Total Environ. 625, 743–751. https://doi.org/10.1016/j.scitotenv.2017.12.324 (2018).ADS 
CAS 
Article 
PubMed 

Google Scholar 
Cao, Y., Zhi, Y., Jeppesen, E. & Li, W. Species-specific responses of submerged macrophytes to simulated extreme precipitation: A mesocosm study. Water 11(6), 1160. https://doi.org/10.3390/w11061160 (2019).CAS 
Article 

Google Scholar 
Hilt, S. et al. Restoration of submerged vegetation in shallow eutrophic lakes—A guideline and state of the art in Germany. Limnologica 36, 155–171 (2006).CAS 
Article 

Google Scholar 
Ryszkowski, L. & Kędziora, A. Modification of water flows and nitrogen fluxes by shelterbelts. Ecol. Eng. 29(4), 388–400. https://doi.org/10.1016/j.ecoleng.2006.09.023 (2007).Article 

Google Scholar  More

(ECHA), E. C. A. Know more about the effects of the chemicals we use in Europe (ECHA/PR/16/01). https://echa.europa.eu/de/-/know-more-about-the-effects-of-the-chemicals-we-use-in-europe (2016).Liu, W. J., Nie, H. X., Liang, D. P., Bai, Y. & Liu, H. W. Phospholipid imaging of zebrafish exposed to fipronil using atmospheric pressure matrix-assisted laser desorption ionization mass spectrometry. Talanta https://doi.org/10.1016/j.talanta.2019.120357 (2020).Article 
PubMed 
PubMed Central 

Google Scholar 
Sparvero, L. J. et al. Mapping of phospholipids by MALDI imaging (MALDI-MSI): Realities and expectations. Chem. Phys. Lipid. 165, 545–562. https://doi.org/10.1016/j.chemphyslip.2012.06.001 (2012).CAS 
Article 

Google Scholar 
Koizumi, S. et al. Imaging mass spectrometry revealed the production of lyso-phosphatidylcholine in the injured ischemic rat brain. Neuroscience 168(1), 219–225. https://doi.org/10.1016/j.neuroscience.2010.03.056 (2010).CAS 
Article 
PubMed 

Google Scholar 
Hankin, J. A. et al. MALDI mass spectrometric imaging of lipids in rat brain injury models. J. Am. Soc. Mass Spectrom. 22(6), 1014–1021. https://doi.org/10.1007/s13361-011-0122-z (2011).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Zhao, C. et al. MALDI-MS imaging reveals asymmetric spatial distribution of lipid metabolites from bisphenol s-induced nephrotoxicity. Anal. Chem. 90(5), 3196–3204. https://doi.org/10.1021/acs.analchem.7b04540 (2018).CAS 
Article 
PubMed 

Google Scholar 
Barbacci, D. C. et al. Mass spectrometric imaging of ceramide biomarkers tracks therapeutic response in traumatic brain injury. ACS Chem. Neurosci. 8(10), 2266–2274. https://doi.org/10.1021/acschemneuro.7b00189 (2017).CAS 
Article 
PubMed 

Google Scholar 
Rompp, A. et al. Histology by mass spectrometry: Label-free tissue characterization obtained from high-accuracy bioanalytical imaging. Angew. Chem. Int. Ed. 49, 3834–3838. https://doi.org/10.1002/anie.200905559 (2010).CAS 
Article 

Google Scholar 
Zemski Berry, K. A. et al. MALDI imaging of lipid biochemistry in tissues by mass spectrometry. Chem. Rev. 111, 6491–6512. https://doi.org/10.1021/cr200280p (2011).CAS 
Article 

Google Scholar 
Cornett, D. S., Reyzer, M. L., Chaurand, P. & Caprioli, R. M. MALDI imaging mass spectrometry: Molecular snapshots of biochemical systems. Nat. Methods 4, 828–833. https://doi.org/10.1038/nmeth1094 (2007).CAS 
Article 
PubMed 

Google Scholar 
Römpp, A. & Spengler, B. Mass spectrometry imaging with high resolution in mass and space. Histochem. Cell Biol. 139, 759–783. https://doi.org/10.1007/s00418-013-1097-6 (2013).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Monroe, E. B. et al. SIMS and MALDI MS imaging of the spinal cord. Proteomics 8(18), 3746-3754. https://doi.org/10.1002/pmic.200800127 (2008).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Chaurand, P., Cornett, D. S., Angel, P. M. & Caprioli, R. M. From whole-body sections down to cellular level, multiscale imaging of phospholipids by MALDI mass spectrometry. Mol. Cell. Proteom. https://doi.org/10.1074/mcp.O110.004259 (2011).Article 

Google Scholar 
Lee, H.-B. & Peart, T. E. Determination of bisphenol A in sewage effluent and sludge by solid-phase and supercritical fluid extraction and gas chromatography/mass spectrometry. J. AOAC Int. 83, 290–298. https://doi.org/10.1093/jaoac/83.2.290 (2000).CAS 
Article 
PubMed 

Google Scholar 
Desbenoit, N., Walch, A., Spengler, B., Brunelle, A. & Römpp, A. Correlative mass spectrometry imaging, applying time-of-flight secondary ion mass spectrometry and atmospheric pressure matrix-assisted laser desorption/ionization to a single tissue section. Rapid Commun. Mass Spectrometry 32, 159–166. https://doi.org/10.1002/rcm.8022 (2018).ADS 
CAS 
Article 

Google Scholar 
Meding, S. et al. Tumor classification of six common cancer types based on proteomic profiling by MALDI imaging. J. Proteome Res. 11, 1996–2003. https://doi.org/10.1021/pr200784p (2012).CAS 
Article 
PubMed 

Google Scholar 
Ritschar, S. et al. Classification of target tissues of Eisenia fetida using sequential multimodal chemical analysis and machine learning. Histochem. Cell Biol. https://doi.org/10.1007/s00418-021-02037-1 (2021).Article 
PubMed 
PubMed Central 

Google Scholar 
Altshuler, I. et al. An integrated multi-disciplinary approach for studying multiple stressors in freshwater ecosystems: Daphnia as a model organism. Integr. Comp. Biol. 51(4), 623–633. https://doi.org/10.1093/icb/icr103 (2011).CAS 
Article 
PubMed 

Google Scholar 
Bambino, K. & Chu, J. in Zebrafish at the Interface of Development and Disease Research Vol. 124 Current Topics in Developmental Biology (ed K. C. Sadler) 331–367 (2017).Seda, J. & Petrusek, A. Daphnia as a model organism in limnology and aquatic biology: Introductory remarks. J. Limnol. 70, 337–344. https://doi.org/10.4081/jlimnol.2011.337 (2011).Article 

Google Scholar 
de Souza Anselmo, C., Sardela, V. F., de Sousa, V. P. & Pereira, H. M. G. Zebrafish (Danio rerio): A valuable tool for predicting the metabolism of xenobiotics in humans? Comp. Biochem. Physiol. Part C: Toxicol. Pharmacol. 212, 34–46. https://doi.org/10.1016/j.cbpc.2018.06.005 (2018).CAS 
Article 

Google Scholar 
Panula, P. et al. The comparative neuroanatomy and neurochemistry of zebrafish CNS systems of relevance to human neuropsychiatric diseases. Neurobiol. Dis. 40, 46–57. https://doi.org/10.1016/j.nbd.2010.05.010 (2010).CAS 
Article 
PubMed 

Google Scholar 
Korn, H. & Faber, D. S. The Mauthner cell half a century later: A neurobiological model for decision-making?. Neuron 47, 13–28. https://doi.org/10.1016/j.neuron.2005.05.019 (2005).CAS 
Article 
PubMed 

Google Scholar 
Schirmer, E., Schuster, S. & Machnik, P. Bisphenols exert detrimental effects on neuronal signaling in mature vertebrate brains. Commun. Biol. https://doi.org/10.1038/s42003-021-01966-w (2021).Article 
PubMed 
PubMed Central 

Google Scholar 
Flößner, D. Book review: Cladocera: The genus Daphnia (including Daphniopsis). Int. Rev. Hydrobiol. 90, 637. https://doi.org/10.1002/iroh.200590003 (2005).Article 

Google Scholar 
OECD. Test No. 211: Daphnia magna Reproduction Test. (2012).Muyssen, B. T. A. & Janssen, C. R. Multigeneration zinc acclimation and tolerance in Daphnia magna: Implications for water-quality guidelines and ecological risk assessment. Environ. Toxicol. Chem. 20, 2053–2060. https://doi.org/10.1002/etc.5620200926 (2001).CAS 
Article 
PubMed 

Google Scholar 
Blewett, T. A. et al. Sublethal and reproductive effects of acute and chronic exposure to flowback and produced water from hydraulic fracturing on the water flea Daphnia magna. Environ. Sci. Technol. 51, 3032–3039. https://doi.org/10.1021/acs.est.6b05179 (2017).ADS 
CAS 
Article 
PubMed 

Google Scholar 
Yang, J. H., Kim, H. J., Lee, S. M., Kim, B. M. & Seo, Y. R. Cadmium-induced biomarkers discovery and comparative network analysis in Daphnia magna. Mol. Cell. Toxicol. 13, 327–336. https://doi.org/10.1007/s13273-017-0036-3 (2017).CAS 
Article 

Google Scholar 
Ferain, A. et al. Body lipid composition modulates acute cadmium toxicity in Daphnia magna adults and juveniles. Chemosphere 205, 328–338. https://doi.org/10.1016/j.chemosphere.2018.04.091 (2018).ADS 
CAS 
Article 
PubMed 

Google Scholar 
Ritschar, S., Narayana, V. K. B., Rabus, M. & Laforsch, C. Uncovering the chemistry behind inducible morphological defences in the crustacean Daphniamagna via micro-Raman spectroscopy. Sci. Rep. 10(1), 22408. https://doi.org/10.1038/s41598-020-79755-4 (2020).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Machnik, P., Schirmer, E., Glück, L. & Schuster, S. Recordings in an integrating central neuron provide a quick way for identifying appropriate anaesthetic use in fish. Sci. Rep. 8, 17541. https://doi.org/10.1038/s41598-018-36130-8 (2018).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Luzio, A. et al. Copper induced upregulation of apoptosis related genes in zebrafish (Danio rerio) gill. Aquat. Toxicol. 128, 183–189. https://doi.org/10.1016/j.aquatox.2012.12.018 (2013).CAS 
Article 
PubMed 

Google Scholar 
Macirella, R. & Brunelli, E. Morphofunctional alterations in zebrafish (Danio rerio) gills after exposure to mercury chloride. Int. J. Mol. Sci. https://doi.org/10.3390/ijms18040824 (2017).Article 
PubMed 
PubMed Central 

Google Scholar 
Mansouri, B. & Johari, S. A. Effects of short-term exposure to sublethal concentrations of silver nanoparticles on histopathology and electron microscope ultrastructure of zebrafish (Danio rerio) gills. IJT 10, 15–20. https://doi.org/10.32598/IJT.10.1.60.4 (2016).CAS 
Article 

Google Scholar 
Perez, C. J., Tata, A., de Campos, M. L., Peng, C. & Ifa, D. R. Monitoring toxic ionic liquids in zebrafish (Danio rerio) with desorption electrospray ionization mass spectrometry imaging (DESI-MSI). J. Am. Soc. Mass Spectrom. 28, 1136–1148. https://doi.org/10.1007/s13361-016-1515-9 (2017).ADS 
CAS 
Article 
PubMed 

Google Scholar 
Stutts, W. L. et al. Methods for cryosectioning and mass spectrometry imaging of whole-body zebrafish. J. Am. Soc. Mass Spectrom. 31, 768–772. https://doi.org/10.1021/jasms.9b00097 (2020).CAS 
Article 
PubMed 

Google Scholar 
Purves, D. & Williams, S. M. Neuroscience. 2nd edition. Vol. Chapter 11, Vision: The Eye (Sinauer Associates, 2001).
Google Scholar 
Strungaru, S. A. et al. Toxicity and chronic effects of deltamethrin exposure on zebrafish (Danio rerio) as a reference model for freshwater fish community. Ecotoxicol. Environ. Saf. 171, 854–862. https://doi.org/10.1016/j.ecoenv.2019.01.057 (2019).CAS 
Article 
PubMed 

Google Scholar 
Mishra, A. & Devi, Y. Histopathological alterations in the brain (optic tectum) of the fresh water teleost Channa punctatus in response to acute and subchronic exposure to the pesticide Chlorpyrifos. Acta Histochem. 116, 176–181. https://doi.org/10.1016/j.acthis.2013.07.001 (2014).CAS 
Article 
PubMed 

Google Scholar 
Jia, W., Mao, L., Zhang, L., Zhang, Y. & Jiang, H. Effects of two strobilurins (azoxystrobin and picoxystrobin) on embryonic development and enzyme activities in juveniles and adult fish livers of zebrafish (Danio rerio). Chemosphere 207, 573–580. https://doi.org/10.1016/j.chemosphere.2018.05.138 (2018).ADS 
CAS 
Article 
PubMed 

Google Scholar 
Seyoum, A., Pradhan, A., Jass, J. & Olsson, P. E. Perfluorinated alkyl substances impede growth, reproduction, lipid metabolism and lifespan in Daphnia magna. Sci. Total Environ. https://doi.org/10.1016/j.scitotenv.2020.139682 (2020).Article 
PubMed 

Google Scholar 
Scanlan, L. D. et al. Gene transcription, metabolite and lipid profiling in eco-indicator Daphnia magna indicate diverse mechanisms of toxicity by legacy and emerging flame-retardants. Environ. Sci. Technol. 49, 7400–7410. https://doi.org/10.1021/acs.est.5b00977 (2015).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Heinlaan, M. et al. Changes in the Daphnia magna midgut upon ingestion of copper oxide nanoparticles: A transmission electron microscopy study. Water Res. 45, 179–190. https://doi.org/10.1016/j.watres.2010.08.026 (2011).CAS 
Article 
PubMed 

Google Scholar 
Abe, T., Saito, H., Niikura, Y., Shigeoka, T. & Nakano, Y. Embryonic development assay with Daphnia magna: Application to toxicity of aniline derivatives. Chemosphere 45, 487–495. https://doi.org/10.1016/s0045-6535(01)00049-2 (2001).ADS 
CAS 
Article 
PubMed 

Google Scholar 
Sengupta, N., Gerard, P. D. & Baldwin, W. S. Perturbations in polar lipids, starvation survival and reproduction following exposure to unsaturated fatty acids or environmental toxicants in Daphnia magna. Chemosphere 144, 2302–2311. https://doi.org/10.1016/j.chemosphere.2015.11.015 (2016).ADS 
CAS 
Article 
PubMed 

Google Scholar 
Huber, K. et al. Approaching cellular resolution and reliable identification in mass spectrometry imaging of tryptic peptides. Anal. Bioanal. Chem. 410, 5825–5837. https://doi.org/10.1007/s00216-018-1199-z (2018).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
White, R. M. et al. Transparent adult zebrafish as a tool for in vivo transplantation analysis. Cell Stem Cell 2, 183–189. https://doi.org/10.1016/j.stem.2007.11.002 (2008).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Nagayoshi, S. et al. Insertional mutagenesis by the Tol2 transposon-mediated enhancer trap approach generated mutations in two developmental genes: tcf7 and synembryn-like. Development 135, 159–169. https://doi.org/10.1242/dev.009050 (2008).CAS 
Article 
PubMed 

Google Scholar 
Perciedu Sert, N. et al. The ARRIVE guidelines 2.0: Updated guidelines for reporting animal research. Exp. Physiol. 105, 1459–1466. https://doi.org/10.1113/EP088870 (2020).Article 

Google Scholar 
Elendt, B. P. Selenium deficiency in Crustacea. Protoplasma 154, 25–33. https://doi.org/10.1007/BF01349532 (1990).CAS 
Article 

Google Scholar 
Sud, M. et al. LMSD: LIPID MAPS structure database. Nucleic Acids Res. 35, D527–D532. https://doi.org/10.1093/nar/gkl838 (2007).CAS 
Article 
PubMed 

Google Scholar 
Race, A. M., Styles, I. B. & Bunch, J. Inclusive sharing of mass spectrometry imaging data requires a converter for all. J. Proteom. 75, 5111–5112. https://doi.org/10.1016/j.jprot.2012.05.035 (2012).CAS 
Article 

Google Scholar 
Robichaud, G., Garrard, K. P., Barry, J. A. & Muddiman, D. C. MSiReader: An open-source interface to view and analyze high resolving power MS imaging files on Matlab platform. J. Am. Soc. Mass Spectrom. 24, 718–721. https://doi.org/10.1007/s13361-013-0607-z (2013).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar  More

Ryther, J. H. Photosynthesis and fish production in the sea. Sci. (80-.) 166, 72–76 (1969).ADS 
CAS 
Article 

Google Scholar 
Follows, M. J., Dutkiewicz, S., Grant, S. & Chisholm, S. W. Emergent biogeography of microbial communities in a model ocean. Sci. (80-.). 315, 1843–1846 (2007).ADS 
CAS 
Article 

Google Scholar 
Edwards, K. F., Litchman, E. & Klausmeier, C. A. Functional traits explain phytoplankton community structure and seasonal dynamics in a marine ecosystem. Ecol. Lett. 16, 56–63 (2013).PubMed 
Article 

Google Scholar 
Nemergut, D. R. et al. Patterns and processes of microbial community assembly. Microbiol. Mol. Biol. Rev. 77, 342–356 (2013).PubMed 
PubMed Central 
Article 

Google Scholar 
Villarino, E. et al. Large-scale ocean connectivity and planktonic body size. Nat. Commun. 9, 142 (2018).Collins, S., Rost, B. & Rynearson, T. A. Evolutionary potential of marine phytoplankton under ocean acidification. Evol. Appl. 7, 140–155 (2014).CAS 
PubMed 
Article 

Google Scholar 
Rusch, D. B. et al. The Sorcerer II global ocean sampling expedition: Northwest Atlantic through Eastern Tropical Pacific. PLOS Biol. 5, e77 (2007).PubMed 
PubMed Central 
Article 
CAS 

Google Scholar 
de Vargas, C. et al. Eukaryotic plankton diversity in the sunlit ocean. Sci. (80-.). 348, 1261605–1/11 (2015).Sunagawa, S. et al. Structure and function of the global ocean microbiome. Sci. (80-.) 348, 1–10 (2015).Article 
CAS 

Google Scholar 
Fuhrman, J. A. et al. A latitudinal diversity gradient in planktonic marine bacteria. Proc. Natl Acad. Sci. 105, 7774–7778 (2008).ADS 
CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Righetti, D., Vogt, M., Gruber, N., Psomas, A. & Zimmermann, N. E. Global pattern of phytoplankton diversity driven by temperature and environmental variability. Sci. Adv. 5, 1–11 (2019).Article 

Google Scholar 
Cermeño, P. et al. The role of nutricline depth in regulating the ocean carbon cycle. PNAS 105, 20344–20349 (2008).ADS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Barton, A. D., Dutkiewicz, S., Flierl, G., Bragg, J. & Follows, M. J. Patterns of diversity in marine phytoplankton. Sci. (80-.) 327, 1509–1511 (2010).ADS 
CAS 
Article 

Google Scholar 
Mantyla, A. W., Venrick, E. L. & Hayward, T. L. Primary production and chlorophyll relationships, derived from ten year of CalCOFI measurements. Calif. Cooperative Ocean. Fish. Investig. Rep. 36, 159–166 (1995).
Google Scholar 
Hayward, T. L. & Venrick, E. L. Nearsurface pattern in the California Current: Coupling between physical and biological structure. Deep. Res. Part II Top. Stud. Oceanogr. https://doi.org/10.1016/S0967-0645(98)80010-6 (1998).Article 

Google Scholar 
Venrick, E. L. Floral patterns in the California Current: The coastal-offshore boundary zone. J. Mar. Res. 67, 89–111 (2009).Article 

Google Scholar 
Powell, J. R. & Ohman, M. D. Covariability of zooplankton gradients with glider-detected density fronts in the Southern California Current System. Deep Sea Res. Part II Top. Stud. Oceanogr. 112, 79–90 (2015).ADS 
CAS 
Article 

Google Scholar 
Taylor, A. G., Landry, M. R., Selph, K. E. & Wokuluk, J. J. Temporal and spatial patterns of microbial community biomass and composition in the Southern California Current Ecosystem. Deep. Res. Part II Top. Stud. Oceanogr. 112, 117–128 (2015).Catlett, D. et al. Diagnosing seasonal to multi-decadal phytoplankton group dynamics in a highly productive coastal ecosystem. Prog. Oceanogr. 197, 102637 (2021).Article 

Google Scholar 
Lilly, L. E. & Ohman, M. D. CCE IV: El Niño-related zooplankton variability in the southern California Current System. Deep. Res. Part I Oceanogr. Res. Pap. 140, 36–51 (2018).ADS 
Article 

Google Scholar 
Richardson, A. J. et al. Using continuous plankton recorder data. Prog. Oceanogr. 68, 27–74 (2006).ADS 
Article 

Google Scholar 
Wang, Z. et al. Microbial communities across nearshore to offshore coastal transects are primarily shaped by distance and temperature. Environ. Microbiol. 1462–2920.14734. https://doi.org/10.1111/1462-2920.14734 (2019).Wang, Y. et al. Patterns and processes of free-living and particle-associated bacterioplankton and archaeaplankton communities in a subtropical river-bay system in South China. Limnol. Oceanogr. 65, S161–S179 (2020).Ibarbalz, F. M. et al. Global Trends in Marine Plankton Diversity across Kingdoms of Life. Cell 1084–1097. https://doi.org/10.1016/j.cell.2019.10.008 (2019).Fuhrman, J. A., Cram, J. A. & Needham, D. M. Marine microbial community dynamics and their ecological interpretation. Nat. Rev. Microbiol. 13, 133–146 (2015).CAS 
PubMed 
Article 

Google Scholar 
Gilbert, J. A. et al. Defining seasonal marine microbial community dynamics. ISME J. 6, 298–308 (2012).CAS 
PubMed 
Article 

Google Scholar 
Karl, D. M. & Lukas, R. The Hawaii Ocean Time-series (HOT) program: background, rationale and field implementation. Deep. Res. Part II Top. Stud. Oceanogr. 43, 129–156 (1996).ADS 
CAS 
Article 

Google Scholar 
Steinberg, D. K. et al. Overview of the US JGOFS Bermuda Atlantic Time-series Study (BATS): A decade-scale look at ocean biology and biogeochemistry Overview of the US JGOFS Bermuda Atlantic Time-series Study (BATS): a decade-scale look at ocean biology and biogeochemistry. Deep. Res. Part II Top. Stud. Oceanogr. 48, 1405–1447 (2015).ADS 
Article 

Google Scholar 
Needham, D. M. & Fuhrman, J. A. Pronounced daily succession of phytoplankton, archaea and bacteria following a spring bloom. Nat. Microbiol. 1, 16005 (2016).Zhu, Z. et al. Understanding the blob bloom: Warming increases toxicity and abundance of the harmful bloom diatom Pseudo-nitzschia in California coastal waters. Harmful Algae 67, 36–43 (2017).CAS 
PubMed 
Article 

Google Scholar 
Mcclatchie, S. et al. State of the California Current 2015–16: Comparisons with the 1997–98 El Niño. Calif. Cooperative Ocean. Fish. Investig. Rep. 57, (2016).Walker, H. J. Jr et al. Unusual occurrences of fishes in the Southern California Current System during the warm water period of 2014–2018. Estuar. Coast. Shelf Sci. 236, 106634 (2020).Article 

Google Scholar 
Kahru, M., Jacox, M. G. & Ohman, M. D. CCE1: Decrease in the frequency of oceanic fronts and surface chlorophyll concentration in the California Current System during the 2014–2016 northeast Pacific warm anomalies. Deep. Res. Part I Oceanogr. Res. Pap. 140, 4–13 (2018).ADS 
Article 

Google Scholar 
Azam, F. et al. The Ecological Role of Water-Column Microbes in the Sea. Mar. Ecol. Prog. Ser. 10, 257–263 (1983).ADS 
Article 

Google Scholar 
Calbet, A. & Landry, M. R. Phytoplankton growth, microzooplankton grazing, and carbon cycling in marine systems. Limnol. Oceanogr. 49, 51–57 (2004).ADS 
CAS 
Article 

Google Scholar 
Buchan, A., LeCleir, G. R., Gulvik, C. A. & González, J. M. Master recyclers: features and functions of bacteria associated with phytoplankton blooms. Nat. Rev. Microbiol. 12, 686–698 (2014).CAS 
PubMed 
Article 

Google Scholar 
Kohonen, T. Exploration of very large databases by self-organizing maps. IEEE Int. Conf. Neural Networks – Conf. Proc. 1, (1997).Istvánovics, V. Eutrophication of Lakes and Reservoirs. Encycl. Inl. Waters 157–165 https://doi.org/10.1016/B978-012370626-3.00141-1 (2009).Partensky, F., Blanchot, J. & Vaulot, D. Differential distribution and ecology of Prochlorococcus and Synechococcus in oceanic waters: a review. Bull. Oceanogr. Monaco 19, 457–475 (1999).
Google Scholar 
Laws, E. A., Falkowski, P. G., Smith, W. O., Ducklow, H. & McCarthy, J. J. Temperature effects on export production in the open ocean. Global Biogeochem. Cycles 14, (2000).Grover, J. P. Resource Competition in a Variable Environment: Phytoplankton Growing According to Monod’s Model. Am. Nat. 136, 771–789 (1990).Article 

Google Scholar 
Benincá, E. et al. Chaos in a long-term experiment with a plankton community. Nature 451, 822–825 (2008).ADS 
PubMed 
Article 
CAS 

Google Scholar 
Williams, R. G. & Follows, M. J. Ocean Dynamics and the Carbon Cycle: Principles and Mechanisms. Book (2011).Lindegren, M., Checkley, D. M., Ohman, M. D., Koslow, J. A. & Goericke, R. Resilience and stability of a pelagic marine ecosystem. Proc. R. Soc. B Biol. Sci. 283, (2016).Vallina, S. M. et al. Global relationship between phytoplankton diversity and productivity in the ocean. Nat. Commun. 1–10 https://doi.org/10.1038/ncomms5299 (2014).Chase, J. M. & Leibold, M. A. Spatial scale dictates the productivity-biodiversity relationship. Nature 416, 427–430 (2002).ADS 
CAS 
PubMed 
Article 

Google Scholar 
Jacox, M. G., Edwards, C. A., Hazen, E. L. & Bograd, S. J. Coastal Upwelling Revisited: Ekman, Bakun, and Improved Upwelling Indices for the U.S. West Coast. J. Geophys. Res. Ocean. 123, 7332–7350 (2018).ADS 
Article 

Google Scholar 
Zaba, K. D. & Rudnick, D. L. The 2014-2015 warming anomaly in the Southern California Current System observed by underwater gliders. Geophys. Res. Lett. 43, 1241–1248 (2016).ADS 
Article 

Google Scholar 
Weber, E. D. et al. State of the California Current 2019–2020: Back to the Future With Marine Heatwaves? Front. Mar. Sci. 8, (2021).Closset, I. et al. Diatom response to alterations in upwelling and nutrient dynamics associated with climate forcing in the California Current System. Limnol. Oceanogr. 1–16. https://doi.org/10.1002/lno.11705 (2021).Kenitz, K. M. et al. Environmental drivers of population variability in colony-forming marine diatoms. Limnol. Oceanogr. 65, 2515–2528 (2020).ADS 
Article 

Google Scholar 
Mullin, M. M. Biomasses of large-celled phytoplankton and their relation to the nitricline and grazing in the California current system off Southern California, 1994–1996. Calif. Cooperative Ocean. Fish. Investig. Rep. 39, 117–123 (1998).
Google Scholar 
Rykaczewski, R. R. & Checkley, D. M. Influence of ocean winds on the pelagic ecosystem in upwelling regions. PNAS 105, 1965–1970 (2007).ADS 
Article 

Google Scholar 
Grzymski, J. J. & Dussaq, A. M. The significance of nitrogen cost minimization in proteomes of marine microorganisms. ISME J. 6, 71–80 (2012).Margalef, R. Life-forms of phytoplankton as survival alternatives in an unstable environment. Ocean. Acta 1, (1978).Falkowski, P. G. & Oliver, M. J. Mix and match: How climate selects phytoplankton. Nat. Rev. Microbiol. 5, 813–819 (2007).Mende, D. R. et al. Environmental drivers of a microbial genomic transition zone in the ocean’s interior. Nat. Microbiol. 2, 1367–1373 (2017).Phoma, B. S. & Makhalanyane, T. P. Depth-dependent variables shape community structure and functionality in the Prince Edward Islands. Microb. Ecol. 81, 396–409 (2021).Kahru, M. & Mitchell, B. G. Seasonal and nonseasonal variability of satellite-derived chlorophyll and colored dissolved organic matter concentration in the California Current. J. Geophys. Res. Ocean. 106, 2517–2529 (2001).ADS 
CAS 
Article 

Google Scholar 
Barth, A., Walter, R. K., Robbins, I. & Pasulka, A. Seasonal and interannual variability of phytoplankton abundance and community composition on the Central Coast of California. Mar. Ecol. Prog. Ser. 637, (2020).Powell, J. R. & Ohman, M. D. Changes in zooplankton habitat, behavior, and acoustic scattering characteristics across glider-resolved fronts in the Southern California Current System. Prog. Oceanogr. 134, 77–92 (2015).ADS 
Article 

Google Scholar 
Taylor, A. G. & Landry, M. R. Phytoplankton biomass and size structure across trophic gradients in the southern California Current and adjacent ocean ecosystems. Mar. Ecol. Prog. Ser. 592, 1–17 (2018).ADS 
CAS 
Article 

Google Scholar 
Dutkiewicz, S., Follows, M. J. & Bragg, J. G. Modeling the coupling of ocean ecology and biogeochemistry. Glob. Biogeochem. Cycles 23, 1–15 (2009).Article 
CAS 

Google Scholar 
D’Ovidio, F., De Monte, S., Alvain, S., Dandonneau, Y. & Lévy, M. Fluid dynamical niches of phytoplankton types. Proc. Natl Acad. Sci. U. S. A. 107, 18366–18370 (2010).ADS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Clayton, S., Dutkiewicz, S., Jahn, O. & Follows, M. J. Dispersal, eddies, and the diversity of marine phytoplankton. Limnol. Oceanogr. Fluids Environ. 3, 182–197 (2013).Article 

Google Scholar 
Moisan, T. A., Rufty, K. M., Moisan, J. R. & Linkswiler, M. A. Satellite observations of phytoplankton functional type spatial distributions, phenology, diversity, and ecotones. Front. Mar. Sci. 4, 1–24 (2017).Article 

Google Scholar 
Combes, V. et al. Cross-shore transport variability in the California Current: Ekman upwelling vs. eddy dynamics. Prog. Oceanogr. 109, 78–89 (2013).ADS 
Article 

Google Scholar 
Chenillat, F., Rivière, P., Capet, X., Franks, P. J. S. & Blanke, B. California coastal upwelling onset variability: cross-shore and bottom-up propagation in the planktonic ecosystem. PLoS ONE 8, (2013).Chenillat, F., Franks, P. J. S. & Combes, V. Biogeochemical properties of eddies in the California Current System. Geophys. Res. Lett. 43, 5812–5820 (2016).ADS 
CAS 
Article 

Google Scholar 
Edwards, K. F., Thomas, M. K., Klausmeier, C. A. & Litchman, E. Allometric scaling and taxonomic variation in nutrient utilization traits and maximum growth rate of phytoplankton. Limnol. Oceanogr. 57, 554–566 (2012).ADS 
Article 

Google Scholar 
Wells, B. K. et al. State of the California Current 2016–17: Still anything but ‘normal’ in the north. Calif. Cooperative Ocean. Fish. Investig. Rep. 58 (2017).Thompson, A. R. et al. State of the California Current 2017–18: Still not quite normal in the north and getting interesting in the south. Calif. Cooperative Ocean. Fish. Investig. Rep. 59 (2018).Ward, C. S. et al. Annual community patterns are driven by seasonal switching between closely related marine bacteria. ISME J. 11, 1412–1422 (2017).Bograd, S. J., Schroeder, I. D. & Jacox, M. G. A water mass history of the Southern California current system. Geophys. Res. Lett. 46, 6690–6698 (2019).ADS 
Article 

Google Scholar 
Parada, A. E., Needham, D. M. & Fuhrman, J. A. Every base matters: Assessing small subunit rRNA primers for marine microbiomes with mock communities, time series and global field samples. Environ. Microbiol. 18 (2016).Amaral-Zettler, L. A., McCliment, E. A., Ducklow, H. W. & Huse, S. M. A method for studying protistan diversity using massively parallel sequencing of V9 hypervariable regions of small-subunit ribosomal RNA Genes. PLoS ONE 4, (2009).Bolyen, E. et al. Reproducible, interactive, scalable and extensible microbiome data science using QIIME 2. Nat. Biotechnol. 37, 852–857 (2019).CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Martin, M. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet.J. 17, (2011).Callahan, B. J., Mcmurdie, P. J., Rosen, M. J., Han, A. W. & A, A. J. DADA2: High resolution sample inference from Illumina amplicon data. Nat. Methods 13, 581–583 (2016).CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Bokulich, N. A. et al. Optimizing taxonomic classification of marker-gene amplicon sequences with QIIME 2’s q2-feature-classifier plugin. Microbiome 6 (2018).Pedregosa, F. et al. Scikit-learn: Machine learning in Python. J. Mach. Learn. Res. 12 (2011).Pruesse, E. et al. SILVA: A comprehensive online resource for quality checked and aligned ribosomal RNA sequence data compatible with ARB. Nucleic Acids Res. 35 (2007).Guillou, L. et al. The Protist Ribosomal Reference database (PR2): A catalog of unicellular eukaryote Small Sub-Unit rRNA sequences with curated taxonomy. Nucleic Acids Res. 41 (2013).McMurdie, P. J. & Holmes, S. Waste Not, Want Not: Why Rarefying Microbiome Data Is Inadmissible. PLoS Comput. Biol. 10 (2014).Gloor, G. B., Wu, J. R., Pawlowsky-Glahn, V. & Egozcue, J. J. It’s all relative: analyzing microbiome data as compositions. Ann. Epidemiol. 26 (2016).Cameron, E. S., Schmidt, P. J., Tremblay, B. J. M., Emelko, M. B. & Müller, K. M. To rarefy or not to rarefy: Enhancing microbial community analysis through next-generation sequencing. bioRxiv. https://doi.org/10.1101/2020.09.09.290049 (2020).Oksanen, J. et al. vegan: Community Ecology Package. R package version 2.5-7. (2020).Bowman, J. S., Amaral-zettler, L. A., Rich, J. J., Luria, C. M. & Ducklow, H. W. Bacterial community segmentation facilitates the prediction of ecosystem function along the coast of the western Antarctic Peninsula. Nat. Publ. Gr. 11, 1460–1471 (2017).
Google Scholar 
Boelaert, J., Bendhaiba, L., Olteanu, M. & Villa-Vialaneix, N. SOMbrero: An R package for numeric and non-numeric self-organizing maps. Adv. Intell. Syst. Comput 295, 219–228 (2014).
Google Scholar 
Johnson, J. B. & Omland, K. S. Model selection in ecology and evolution. Trends Ecol. Evol. 19, 101–108 (2004).PubMed 
Article 

Google Scholar 
James, C. C. et al. Influence of nutrient supply on plankton microbiome biodiversity and distribution in a coastal upwelling region. https://doi.org/10.5281/zenodo.6359865 (2022).Legendre, P. & Legendre, L. Numerical ecology (Elsevier, 2012). More

Overall strategyAn admixture analysis aims to estimate the admixture proportions (or ancestries), Q, of each sampled individual in a given number of K source populations (Pritchard et al. 2000), and the characteristic allele frequencies, P, at each locus of each inferred source population. Even though Q is frequently of the primary interest, P must be estimated simultaneously because we have genotype data only and Q is highly dependent on P which actually defines the source populations. For N individuals from K source populations genotyped at L loci with a total number of A alleles, the numbers of independent variables in Q and P are VQ = (K − 1)N and VP = (A − L)K, respectively. The high dimensionality of an admixture analysis, with V = VQ + VP = (K − 1)N + (A − L)K variables, not only incurs a large computational burden, but also poses a high risk of non-convergence (to the global maximum) for any algorithm, especially when either Q or P is expected to be poorly estimated in difficult situations such as a small sample (say, a couple) of individuals from each source population or low differentiation.I propose a two-step procedure with corresponding algorithms to reduce the risk of non-convergence, to speed up the computation, and to make more accurate inferences of both Q and P. In the first step, I assume a mixture model (Pritchard et al. 2000; Falush et al. 2003) that individuals in a sample can come from different source populations, but each individual’s genome comes exclusively from a single population. Under this simplified probabilistic model, I conduct a clustering analysis to obtain estimates of both individual memberships and allele frequencies of each cluster by a global maximisation algorithm, simulated annealing, with extra care (details below) of convergence. In the absence of admixture and with sufficient information for complete recovery of population structure, the estimated individual memberships and allele frequencies of the clusters are expected to be equivalent to Q (with element qik = 1 and qil = 0 if individual i is inferred to be in cluster k where l ≠ k) and P, respectively. Otherwise, they are expected to be good approximations of Q and P, because an admixed individual i with the highest ancestral proportion from a population would be expected to be assigned (exclusively) to that population. In the second step, I assume an admixture model (Pritchard et al. 2000; Falush et al. 2003) to refine estimates of Q and P, using an EM algorithm and the start parameter (Q and P) values obtained from the clustering analysis. Because the starting values are already close to the truth, the algorithm is fast and has a much-reduced risk of converging to a local maximum than the original EM algorithms (Tang et al. 2005; Alexander et al. 2009).Clustering analysisI assume N diploid individuals are sampled from K source populations. The origin of a sampled individual from the K source populations is unknown, which is the primary interest of structure analysis. However, if it is (partially) known, this information can be used to supervise (help) the clustering analysis of other sampled individuals of unknown origins. Each individual’s genome comes exclusively from one of the K unknown source populations (i.e., mixture model, no admixture). I assume each individual is genotyped at L loci, with a diploid genotype {xil1, xil2} for individual i (=1, 2, …, N) at locus l (=1, 2, …, L). The task of the clustering analysis is to sort the N individuals with genotype data X = {xila:i = 1, 2, …, N; l = 1, 2, …, L; a = 1, 2} into K clusters, with each representing a source population. No assumption is made about the evolutionary relationships of the populations, which, when summarized by F statistics, are estimated from the same genotype data in both clustering and admixture analyses.Suppose, in a given clustering configuration Ω = {Ω1, Ω2, …, ΩK}, cluster k (=1, 2, …, K), Ωk, contains a set of Nk (with Nk  > 0 and (mathop {sum}nolimits_{k = 1}^K {N_k equiv N})) individuals, denoted by Ωk = {ωk1, ωk2, …, ωkNk} where ωkj is the index of the jth individual in cluster k. The genotype data of the Nk individuals in cluster k is Xk = {xila: i ∈ Ωk; l = 1, 2, …, L; a = 1, 2}. The log-likelihood of Ωk is then the log probability of observing Xk given Ωk$${{{mathcal{L}}}}_kleft( {{{{mathbf{Omega }}}}_k} right) = {{{mathrm{LogP}}}}left( {{{{mathbf{X}}}}_kleft| {{{{mathbf{Omega }}}}_k} right.} right) = mathop {sum}limits_{l = 1}^L {mathop {sum}limits_{j = 1}^{J_l} {c_{klj}{{{mathrm{Log}}}}left( {p_{klj}} right)} }$$
(1)
where cklj and pklj are the count of copies and the frequency, respectively, of allele j at locus l in cluster k, and Jl is the number of alleles at locus l. Given Ωk, cklj is counted from genotype data Xk, and allele frequency pklj is estimated by$$p_{klj} = left( {p_{lj} + c_{klj}} right)/mathop {sum}limits_{m = 1}^{J_l} {left( {p_{lm} + c_{klm}} right)}$$
(2)
where plj is the frequency of allele j at locus l in the entire population represented by the K clusters. plj is calculated by$$p_{lj} = mathop {sum}limits_{k = 1}^K {c_{klj}} /mathop {sum}limits_{m = 1}^{J_l} {mathop {sum}limits_{k = 1}^K {c_{klm}} } = c_{lj}/mathop {sum}limits_{m = 1}^{J_l} {c_{lm}}$$
(3)
where (c_{lm} = mathop {sum}nolimits_{k = 1}^K {c_{klm}}) is the count of allele m (=1, 2, …, Jl) at locus l in the entire sample of individuals.Under the mixture model above, clusters are only weakly dependent (with the extent of dependency decreasing with an increasing value of K) and the total log-likelihood of the clustering configuration, Ω = {Ω1, Ω2, …, ΩK}, is thus$${{{mathcal{L}}}}left( {{{mathbf{Omega }}}} right) = mathop {sum}limits_{k = 1}^K {{{{mathcal{L}}}}_kleft( {{{{mathbf{Omega }}}}_k} right)} ,$$
(4)
where ({{{mathcal{L}}}}_kleft( {{{{mathbf{Omega }}}}_k} right)) is calculated by (1).It is worth noting that allele frequencies, P, are modelled as hidden or nuisance variables and are estimated as a by-product of maximising (4) for estimates of Ω. Yet, careful modelling of P proves important for estimating Ω, as the two are highly dependent. Bayesian admixture methods assume allele frequencies pkl = {pkl1, pkl2, …, (p_{klj_l})} in a Dirichlet distribution (e.g., Foreman et al. 1997; Rannala and Mountain 1997; Pritchard et al. 2000), ({{{mathcal{D}}}}left( {lambda _1,lambda _2, ldots ,lambda _{J_l}} right)). For any population k, the uncorrelated (Pritchard et al. 2000) and correlated (Falush et al. 2003) allele frequency model assumes λj = 1 and (lambda_j=p_{ol_j}F_K/(1-F_k)), respectively, for j = 1, 2, …, Jl. In the latter model, p0lj is the frequency of allele j at locus l in the ancestral population (common to the K derived populations), and Fk is the differentiation of population k from the ancestral population. In contrast, likelihood admixture methods (e.g., Tang et al. 2005; Alexander et al. 2009; Frichot et al. 2014) and non-model based clustering methods (e.g., K-means method, Jombart et al. 2010) do not use any prior, which is equivalent to assuming plj ≡ 0 for j = 1, 2, …, Jl in Eq. (2). However, properly modelling prior allele frequencies, as carefully considered in Bayesian methods (Pritchard et al. 2000; Falush et al. 2003), becomes important in situations where allele frequencies are not well defined or tricky to estimate, such as when few individuals are sampled from a source population or when rare alleles are present. The frequentist estimator (2) is in spirit similar to the Bayesian correlated allele frequency model (Falush et al. 2003), and leads to accurate results in various situations to be shown in this study. I have also tried alternatives such as plj ≡ 1/Jl (which is similar to the uncorrelated allele frequency model of Pritchard et al. 2000) or plj ≡ 0 (which is equivalent to the treatment in previous likelihood admixture analysis or non-model based clustering analysis) in replacement of (2), but none works as well as (2) and could yield much less accurate results in difficult situations (below).Scaling for unbalanced samplingBayesian methods of STRUCTURE’s admixture model assume an individual i’s ancestry, qi = {qi1, qi2, …, qiK}, follows a prior Dirichlet probability distribution ({{{mathbf{q}}}}_isim {{{mathbf{{{{mathcal{D}}}}}}}}left( {alpha _1,alpha _2, ldots ,alpha _K} right)) (Pritchard et al. 2000; Falush et al. 2003). By default, α1 = α2 = ··· = αK = α, which essentially assumes that an individual has its ancestry originating from each of the assumed K populations at an equal prior probability of 1/K. To model unequal sample sizes such that an individual comes from a more intensively sampled population at a higher prior probability, STRUCTURE also has applied an alternative prior, α1 ≠ α2 ≠ ··· ≠ αK. It is shown that, when sampling intensity is heavily unbalanced among populations, the default prior could lead to the split of a large cluster and the merge of small clusters, while the alternative prior yields much more accurate results (Wang 2017). These priors have a large impact on admixture analysis; applying the default prior to data of highly unbalanced samples leads to inaccurate Q estimates even when many informative markers are used (Wang 2017).Unfortunately, current non-model based or likelihood-based admixture analysis methods do not utilise these or other priors for handling unbalanced sampling. As a result, they can give inaccurate admixture estimates, just like STRUCTURE under the default ancestry prior model, for data from highly unbalanced sampling. To reduce the cluster split and merge problems, herein I propose the following method to scale the likelihood of a cluster by the size, the number of individual members, of the cluster.The original log-likelihood of cluster k, ({{{mathcal{L}}}}_kleft( {{{{mathbf{Omega }}}}_k} right)), is calculated by (1). It is then scaled by the cluster size, Nk, as$${{{mathcal{L}}}}_{Sk}left( {{{{mathbf{Omega }}}}_k} right) = {{{mathcal{L}}}}_kleft( {{{{mathbf{Omega }}}}_k} right)/left( {1 + e^{sN_k/left( {8N} right)}} right),$$
(5)
where s is the scaling factor taking values 1, 2, 3 for weak, medium and strong scaling, respectively. This scaling scheme encourages large clusters and discourages small clusters. Although (5) is not an analytically derived but an empirical equation and is thus not guaranteed to be optimal, extensive simulations (some shown below) verify that the scaling scheme works very well for data from highly unbalanced sampling, yielding accurate clustering analysis results and thus similarly or more accurate admixture estimates than STRUCTURE under its alternative ancestry model. The most appropriate scaling level (1, 2 or 3) for a particular dataset depends on how unbalanced the sampling is, how much differentiated the populations are, and how much informative the markers are. For example, a low scaling level, s = 1, is appropriate when many markers are genotyped for a set of lowly differentiated (low FST) populations. Usually, we do not know these factors in analysing the data. Therefore, when the data are suspected to be unbalanced in sampling among populations, they are better analysed comparatively with different levels of scaling (0, 1, 2, and 3). When the applied level of scaling is too low, large populations tend to be split and small populations tend to be merged. When the applied level of scaling is too high, small populations tend to be merged among themselves or with a large population. With the help of some internal information such as consistency of replicate runs at the same scaling level and the same K value and some external information such as sampling locations in examining the admixture estimates, the appropriate scaling level can be determined.Simulated annealing algorithmA likelihood function with many variables, such as (4), is difficult to maximise for estimates of the variables. Traditional methods, such as derivative based Newton-Raphson algorithm (e.g., Tang et al. 2005) and non-derivative based EM algorithm (Dempster et al. 1977; Tang et al. 2005; Alexander et al. 2009), may converge to a local rather than the global maximum for a large scale problem with ridges and plateaus (Gaffe et al. 1994). Although trying multiple replicate runs with different starting values and choosing the run with the highest likelihood could reduce the risk of landing on a local maximum, a global maximum cannot be guaranteed regardless of the number of runs. The Bayesian approach as implemented in STRUCTURE (Pritchard et al. 2000) has a similar problem, as different replicate runs of the same data with the same parameter and model choices but different random number seeds may yield different admixture estimates and likelihood values (Tang et al. 2005; below).Simulated annealing (SA) was developed to optimise very large and complex systems (Kirkpatrick et al. 1983). Using the Metropolis algorithm (Metropolis et al. 1953) from statistical mechanics, SA can find the global maximum by searching both downhill and uphill and by traversing deep valleys on the likelihood surface to avoid getting stuck on a local maximum (Kirkpatrick et al. 1983; Goffe et al. 1994). It has been proved to be highly powerful in pedigree reconstruction (Wang 2004; Wang and Santure 2009) from genotype data, which is probably more difficult than population structure reconstruction (i.e., clustering analysis) because the genetic structure (i.e., sibship) of the former is, in general, more numerous, more complicated with hierarchy, and smaller (thus more elusive and more difficult to define) than that in the latter. Herein I propose a SA algorithm for a population clustering analysis, as detailed in Supplementary Appendix 1.Admixture analysisUnder the mixture model, the above clustering analysis partitions the N sampled individuals into a predefined K clusters, each representing a source population. The properties (e.g., genetic diversity) of and the relationships (e.g., FST) among these populations can be learnt from the inferred clusters. However, the clustering results are accurate only when the mixture model is valid. For a sample containing a substantial proportion of highly admixed individuals (i.e., who have recent ancestors from multiple source populations), the clustering results are just approximations. In such a case, the admixture model is more appropriate and can be used to refine the mixture analysis results by inferring the admixture proportions (or ancestry coefficients) of each sampled individual.Under the admixture model (Pritchard et al. 2000), an individual i’s ancestry (or admixture proportions) can be characterised by a vector qi = {qi1, qi2, …, qiK}, where qik is the proportion of its genome coming from (contributed by) source population k. Equivalently, qik can also be taken as the probability that an allele sampled at random from individual i comes from source population k. Obviously, we have qik ≥ 0 and (mathop {sum}nolimits_{k = 1}^K {q_{ik} equiv 1}). The overall admixture extent of individual i can be measured by (M_i = 1 – mathop {sum}nolimits_{k = 1}^K {q_{ik}^2}), the probability that the two alleles at a randomly drawn locus come from different source populations. Individual i is purebred and admixed when Mi = 0 and Mi  > 0, respectively. An F1 and F2 hybrid individual i is expected to have Mi = 0.5 and Mi = 0.625, respectively.The task of an admixture analysis is to infer qi for each individual i, denoted by Q = {q1, q2, …, qN}. The log-likelihood function is$${{{mathcal{L}}}}left( {{{{mathbf{Q}}}},{{{mathbf{P}}}}left| {{{mathbf{X}}}} right.} right) = mathop {sum}limits_{i = 1}^N {mathop {sum}limits_{l = 1}^L {mathop {sum}limits_{a = 1}^2 {{{{mathrm{Log}}}}left( {mathop {sum}limits_{k = 1}^K {q_{ik}p_{klx_{ila}}} } right)} } }$$
(6)
Note (6) is essentially the same as those proposed in previous studies (e.g., Tang et al. 2005; Alexander et al. 2009). It assumes independence of individuals conditional on the genetic structure defined by Q, and independence of alleles both within and between loci. The former can be violated when the data have genetic structure in addition to the subpopulation structure defined by Q, such as the presence of familial structure (Rodríguez‐Ramilo and Wang 2012) or inbreeding (Gao et al. 2007) within a subpopulation. The assumption of independence among loci is violated for markers in linkage disequilibrium. It, as well as the assumption of independence between paternal and maternal alleles within a locus, is also violated due to admixture (Tang et al. 2005) or inbreeding (Gao et al. 2007). However, (6) is a good approximation and works well in general even when these assumptions are violated, as checked by extensive simulations.If P were known, it would be trivial to estimate Q from X. Unfortunately, usually, the only information we have is genotype data X, from which we must infer K, Q and P jointly. Herein I modify the EM algorithm of Tang et al. (2005) to solve (6) for maximum likelihood estimates of Q and P given K, as detailed in Supplementary Appendix 2.Despite essentially the same likelihood function, my EM algorithm differs from that of Tang et al. (2005) in three aspects. First, I use the clustering results of mixture model as initial values of Q. Even in the worst scenario of many highly admixed individuals included in a sample, the clustering results should still be much closer to the true Q than a random guess, as used in previous likelihood methods (Tang et al. 2005; Alexander et al. 2009). It is possible (and indeed it has been trialled) to use the results of a faster non-model based clustering method, such as K-means method, in place of those of the likelihood-based clustering method with simulated annealing algorithm as described above. However, such non-model based methods are less reliable and less accurate, especially in difficult situations (below). Second, rather than updating Q and P in alternation, I update Q to asymptotic convergence under a given P. I then update P using the converged Q. This two-step iteration process is repeated until the convergence of both Q and P is reached. Third, the allele frequencies for a specific individual i are calculated by excluding the genotypes of the individual, which are then used in the EM procedure for iteratively updating qi.Optimal KThe above-described clustering analysis and admixture analysis are conducted by assuming a given number of source populations, K. Apparently, different genetic structures would be inferred from the same genotype data if different K values are assumed. In some cases, a reasonable K value is roughly known. For example, individuals might be sampled from K known discrete locations (say, lakes), and the purposes of a structure analysis are to confirm that populations from different locations are indeed differentiated and thus distinguishable, to identify migrants between the locations, and to find out the patterns of genetic differentiations (e.g., whether isolation by distance applies or not). In many other cases, however, we may have no idea of the most likely K value. For example, individuals might be sampled from the same breeding or feeding ground and we wish to know how many populations are using the same ground, and to learn the properties of these populations from the individuals sampled and assigned to them. In such a situation of hidden genetic structure, we need first to identify the most likely one or more K values, and then investigate the corresponding structure/admixture.Estimating the most likely K value from genotype data is difficult (Pritchard et al. 2000). Although many methods have been proposed and applied (see review by Wang 2019), they are all ad hoc to some extent and may be inaccurate in difficult situations such as highly unbalanced sampling from different populations and low differentiation (Wang 2019). Herein I propose two ad hoc estimators of K that can be calculated from the clustering analysis presented in this study. They have a satisfactory accuracy as checked by many test datasets, simulated and empirical.The first estimator is based on the second order rate of change of the estimated log-likelihood as a function of K in a clustering analysis, DLK2. This estimator is similar in spirit to the ∆K method of Evanno et al. (2005), but does not use the mean and standard deviation of log-likelihood values among replicate runs (for a given K value) because the standard deviation (the denominator of ∆K) is frequently zero thanks to the convergence of our clustering analysis by the simulated annealing algorithm.The second estimator, denoted by FSTIS, is based on Wright (1984)’s F-statistics. The best K should produce the strongest population structure, with high differentiation (measured by FST) of each inferred cluster and low deviation from Hardy-Weinberg equilibrium (measured by FIS) within each inferred cluster. Details of how to calculate the two estimators are in Supplementary Appendix 3.SimulationsTo evaluate the accuracy, robustness, and computational efficiency of the new methods implemented in PopCluster in comparison with other methods, I simulated and analysed data with different population structures and sampling intensities. The simulation procedure described below is implemented in the software package PopCluster.Simulation 1, small samplesA population becomes difficult to define genetically when few individuals from it are sampled and included in an admixture analysis. However, a small sample of individuals can be common in practice when, for example, archaeological samples (usually few) are used in studying ancient population structure or in studying the relationship between ancient and current populations (e.g., Lazaridis et al. 2014). In a mixed stock analysis (Smouse et al. 1990) or a wildlife forensic analysis of source populations, there might also be few sampled individuals representing a rare population. To investigate the impact of sample sizes on an admixture analysis, I simulated 10 populations in an island model with FST = 0.05. Nk (=2, 3, …, 10 and 20) individuals were sampled from each of the 10 populations, or 1 individual was sampled from each of the first five populations and 2 individuals were sampled from each of the last five populations (the case Nk = 1.5, Table 1). Other simulation parameters are summarized in Table 1.Table 1 Simulation parameters.Full size tableSimulation 2, many populationsAdmixture becomes increasingly difficult to infer with an increasing K, the number of assumed populations, because the dimensions of both Q and P increase linearly with K. This contrasts with the number of individuals, N, and the number of loci, L, which determines the dimensions of Q and P only, respectively. Therefore, the scale of an admixture analysis, in terms of the number of parameters to be estimated, is predominantly determined by K rather than N or L. I simulated data with a widely variable number of populations (K = [6, 100]) to see if the structure can be accurately reconstructed by using relatively highly informative markers (parameters in Table 1), especially when K is large which is rarely considered in previous simulation studies.Simulation 3, spatial admixture modelThe spatial admixture model resembles isolation by distance where population structure changes gradually as a function of geographic location. Under this model, populations are not discrete as assumed by admixture models and have no recognisable boundaries, posing challenges to an admixture analysis. To simulate the spatially gradual changes in genetic structure, I assume source populations 1, 2, …, K are equally spaced in that order along a line (say, a river in reality). Sampled individuals 1, 2, …, N are also equally spaced in that order on the same line. The admixture proportions of individual i, qi = {qi1, qi2, …, qiK}, being the proportional genetic contributions to i from source populations k, are a function of the individual’s proximity to these K source populations. Formally, we have$$q_{ik} = frac{{q_{ik}^ ast }}{{mathop {sum}nolimits_{k = 1}^K {q_{ik}^ ast } }}$$
(7)
where$$q_{ik}^ ast = left[ {1 – left( {frac{{i – 1}}{{N – 1}} – frac{{k – 1}}{{K – 1}}} right)^2} right]^S$$and parameter S is used to regulate the admixture extent of the N sampled individuals. Under this spatial admixture model, an individual i’s admixture (qi) is determined by its location, or the distances from the K source populations. The 1st and the last sampled individuals (i = 1, N) always have the least admixture, measured by (M_i = 1 – mathop {sum}nolimits_{k = 1}^K {q_{ik}^2}). q11 (=qNK) is always the largest among the qik values for i = 1, 2, …, N and k = 1, 2, …, K. Given a desired value of q11 and K, the scaler parameter S can be solved from the above equations. Given K, N and S, qi of an individual i can then be calculated from the above equations. In this study, I simulated and analysed samples generated with parameters K = 5, N = 500, L = 10000 SNPs, and q11 varying between 0.5 and 1.0 (Table 1).Simulation 4, low differentiationPopulation structure analysis becomes increasingly difficult with a decreasing differentiation, usually measured by FST, among subpopulations. Fortunately, with genomic data of many SNPs, it is still possible to detect weak and subtle population structures (Patterson et al. 2006) as demonstrated in human fine-structure analysis (e.g., Leslie et al. 2015). I simulated data with varying weak population structures (low FST, Table 1) and otherwise ideal populational (only 3 equally differentiated subpopulations) and sampling conditions (i.e., a large sample of individuals per subpopulation, and many SNPs). The number of SNPs used in analyse was L = 1000/FST such that in principle the population structures should be inferred with roughly equal power and accuracy. Because L is large for low FST, STRUCTURE analysis was abandoned due to computational difficulties.Simulation 5, unbalanced samplingSamples of individuals from different source populations are rarely identical in size in practice. Frequently, different source populations are represented by different numbers of individuals in a sample. The impact of unbalanced sampling and how to mitigate it in applying STRUCTURE have been investigated (e.g., Puechmaille 2016; Wang 2017). Similar problems exist for other admixture or clustering analysis methods but have not been studied yet. The same population structure and unbalanced sampling schemes (see parameters in Table 1) used in Wang (2017) were used to simulate data, which were then analysed by various methods to understand their robustness to unbalanced sampling.Simulation 6, computational efficiencySamples from a variable number of populations (Table 1) were analysed by the four programs on a linux cluster to compare their computational efficiencies. Each program uses a single core (no parallelisation) of a processor (Intel Xeon Gold 6248 2.5 GHz) for a maximal allowed time of 48 or 72 (when K = 1024 only) hours. Default parameter settings are used for all four programs. For STRUCTURE, both burn-in and run lengths were set to 104, although much higher burn-in is required for convergence when K is large (say K  > 20). The running time for STRUCTURE is thus conservative, especially when K is not small.Further simulations were conducted to investigate the effects of high admixture and the presence of familial relationships and inbreeding on the relative performance of different admixture analysis methods, as detailed in Supplementary Appendix 4.In all simulations except for the spatial admixture model, I assumed a population with K discrete subpopulations in Wright’s (1931) island model in equilibrium among mutation, drift and migration. For a locus l (=1, 2, …, L) with Jl alleles, allele frequencies of the ancestral population, p0l = {p0l1, p0l2, …, (p_{0lJ_l})}, were drawn from a uniform Dirichlet distribution, ({{{mathcal{D}}}}left( {lambda _1,lambda _2, ldots ,lambda _{J_l}} right)) where λj = 1 for j = 1, 2, …, Jl. Given p0l, allele frequencies of subpopulation k (=1, 2, …, K), pkl = {pkl1, pkl2, …, (p_{klJ_l})}, were drawn from a uniform Dirichlet distribution, ({{{mathcal{D}}}}left( {lambda _1,lambda _2, ldots ,lambda _{J_l}} right)), where (lambda _j = ( {frac{1}{{F_{ST}}} – 1} )p_{0lj}) for j = 1, 2, …, Jl (Nicholson et al. 2002; Falush et al. 2003). Given pkl and the admixture proportion qi of individual i, two alleles at locus l were drawn independently to form the individual’s genotype. The multilocus genotype of an individual was obtained by combining single locus genotypes sampled independently, assuming linkage equilibrium. Nk individuals were drawn at random from population k (= 1, 2, …, K), which were then pooled and subjected to a structure analysis.For the spatial population and sampling model, allele frequencies at a locus l, p0l and pkl, are generated as before, assuming FST = 0.05 among K = 5 subpopulations. A number of N = 500 individuals, equally spaced on the line between source populations 1 and 5, are sampled. The admixture proportion of individual i, qi, is determined by its location, calculated by Eq. (7). Given pkl and qi, the multilocus genotype of individual i is simulated as described above.For each parameter combination, 100 replicate datasets were simulated, analysed and assessed for estimation accuracy. Each dataset was analysed for admixture by different methods (see below for details) with an assumed K as used in simulations. I did not consider estimating the optimal K by analysing a simulated dataset in a range of possible K values. This is because, like previous studies (e.g., Pritchard et al. 2000; Alexander et al. 2009), I am more concerned with admixture inference under a given K, which is important of itself and forms the basis for inferring the optimal K as well. This is also because it is almost impossible computationally to estimate the optimal K for so many replicate datasets and so many parameter combinations in a large-scale simulation study like the present one, even when using large computer clusters. The optimal K was estimated for several empirical datasets (below).Measurement of accuracyInference accuracy could be assessed by comparing, for each individual i, the agreement between simulated ancestry coefficients, qi, and estimated ancestry coefficients, (widehat {{{mathbf{q}}}}_i), obtained by an admixture analysis assuming the true/simulated subpopulation number K. Because the reconstructed populations are labelled arbitrarily (Pritchard et al. 2000), no meaningful results can be gained by comparing qi and (widehat {{{mathbf{q}}}}_i) directly, however. It is possible to relabel the reconstructed populations and find the labelling scheme that has the maximum agreement between qi and (widehat {{{mathbf{q}}}}_i) as the measurement of accuracy. However, there are K! possible labelling schemes, making the approach difficult to calculate when K is large (say, K > 50).The labelling becomes irrelevant when pairs of individuals are considered for the co-assignment probabilities (or coancestry) (Dawson and Belkhir 2001). I calculate and use the average difference between simulated and estimated coancestry for pairs of sampled individuals to measure the average assignment error, AAE (Wang 2017),$$AAE = left( {frac{1}{{Nleft( {N – 1} right)/2}}mathop {sum}limits_{i = 1}^N {mathop {sum}limits_{j = 1 + 1}^N {left( {mathop {sum}limits_{k = 1}^K {q_{ik}q_{jk}} – mathop {sum}limits_{k = 1}^K {widehat q_{ik}widehat q_{jk}} } right)^2} } } right)^{1/2}.$$
(8)
The minimum value of AAE is 0, when ancestry (admixture) is inferred perfectly. The maximum value is 1, when there are no admixed individuals in the sample, individuals from the same source population are always assigned to different populations and individuals from different source populations are always assigned to the same population. It is worth noting that the minimum AAE value of 0 is always possible for any population structure. However, the maximum value varies and can be much smaller than 1, depending on the actual underlying population structure. With an increasing K value or increasing admixture (i.e., qik→1/K for any individual i), the maximum value of AAE tends to decrease. For this reason, AAE cannot be compared fairly between different genetic structures (e.g., different K values, different actual Q for a given K, or different sizes of subsamples from the source populations) for measuring the relative inference qualities. However, it can always be used to compare the accuracy of different inference methods for a given simulated genetic structure and a given sample.Analysis of real datasetsAn ant datasetIt was originally used in a study of the mating system of an ant species, Leptothorax acervorum (Hammond et al. 2001). Ten sampled colonies, A, B, C, D, E, F, G, H, I, and J, contribute respectively 9, 7, 47, 45, 45, 45, 45, 45, 44, and 45 diploid workers to a sample of 377 individuals. For this species, we know that each colony is headed by a single diploid queen mated with a single haploid male. Therefore, workers from the same colony are full-sibs and workers from different colonies are non-sibs. Each sampled worker was genotyped at up to 6 microsatellite loci, which have 3 to 22 alleles per locus observed in the 377 individuals. This dataset was analysed to reconstruct the genetic structure of the sample, which actually is the family structure. ADMIXTURE and sNMF cannot handle multiallelic marker data and therefore only STRUCTURE and PopCluster are used for analysing this dataset.For STRUCTURE, I used the default parameter settings, except for the burning-in and run lengths which were both set to 105 to reduce the risk of non-convergence. Two analyses were conducted. First, optimal K values were determined using three estimators (Wang 2019) calculated from STRUCTURE outputs, and using the DLK2 estimator of PopCluster. For this K estimation purpose, 20 replicate runs for each possible K value in the range [1, 15] were conducted by both STRUCTURE and PopCluster. Second, assuming K = 10, a number of 100 replicate runs (each with a distinctive seed for the random number generator) were conducted by both STRUCTURE and PopCluster to investigate their convergence.An Arctic charr datasetShikano et al. (2015) sampled 328 Arctic charr individuals from 6 locations in northern Fennoscandia: two lakes (Galggojavri and Gallajavri) and one pond (Leenanlampi) in the Skibotn watercourse drain into the Atlantic Ocean and three lakes (Somasjärvi, Urtas-Riimmajärvi and Kilpisjärvi) in the Tornio-Muoniojoki watercourse drain into the Baltic Sea. Individuals were genotyped at 15 microsatellite loci to study the genetic structure and demography. The data were again analysed by STRUCTURE and PopCluster but not by ADMIXTURE and sNMF because the markers are multiallelic. I conducted two separate analyses of the genotype data. First, I estimated the most likely K value by each program, making 20 replicate runs with each K value in the range [1, 10]. Second, I investigated the convergence of each program by conducting 100 replicate runs of the data at K = 6. STRUCTURE analyses were run with default parameter settings except for both burn-in and run lengths being 105.A human SNP datasetUsing FRAPPE (Tang et al. 2005), Li et al. (2008) studied the world-wide human population structure represented by 938 individuals sampled from 51 populations of the Human Genome Diversity Panel (HGDP). Each individual was genotyped at 650000 common SNP loci. The data were expanded to include genotypes of 1043 individuals at 644258 SNPs, available from http://www.cephb.fr/en/hgdp_panel.php#basedonnees. In this study, the expanded data were comparatively analysed by PopCluster, ADMIXTURE, and sNMF, assuming K = 7 clusters (regions) as in the original study (Li et al. 2008). STRUCTURE was too slow to analyse this big dataset and thus it was abandoned.The human 1000 genomes phase I datasetThe dataset (Abecasis et al. 2012), available from https://www.internationalgenome.org/data/, has 1092 human individuals sampled from 14 populations across all continents, with each individual having 38 million SNP genotypes. After removing monomorphic loci (note, no pruning was applied regarding missing data, minor allele frequency and linkage disequilibrium, in contrast to other studies), genotypes at a number of L = 38035992 SNPs were analysed by PopCluster and sNMF, assuming K = 9 clusters (regions). Both STRUCTURE and ADMIXTURE were too slow to analyse this huge dataset and thus were abandoned. No attempts are made to find the optimal K for this dataset as done for the ant and Arctic charr datasets, because too much computational time is required for PopCluster or sNMF to analyse the data with a number of replicate runs at each of a number of K values even when using a large cluster, and there might be multiple K values that explain the data equally well (at different spatial and time scales). For a better understanding of the world-wide human population genetic structure, the data should be analysed at least with one replicate under each of a number of possible K values, say K = [5, 12], to reveal and compare the genetic structure. This study analysed the data at a single K = 9 for the purpose of demonstrating the capacity of different methods, and comparing the admixture estimates of PopCluster and sNMF at this particular value of K. Because of the incompleteness of the analysis, the biological interpretations of the results should be taken with caution.Comparative analyses by different softwareI compared the accuracy and computational time of STRUCTURE (Pritchard et al. 2000; Falush et al. 2003), ADMIXTURE (Alexander et al. 2009), sNMF (Frichot et al. 2014) and PopCluster in analysing both simulated and empirical datasets described above. Quite a few other model-based methods implemented in various software exist. I choose STRUCTURE and ADMIXTURE because they are the most popular model-based admixture analysis methods used for small and large datasets, respectively. I also choose sNMF because it is a very fast model-based method that works for huge datasets for which other methods, such as ADMIXTURE, fail to run or take unrealistically too much time to run.STRUCTURE can handle both diallelic (such as SNPs) and multiallelic (such as microsatellites) markers, but runs too slowly to analyse large datasets with many markers, many individuals, or many populations. It was therefore used to analyse all simulated and empirical datasets with no more than 10000 loci. The default parameter setting was used for most datasets, with a burn-in length of 104 and a run length of 104 iterations. For better convergence, the burn-in and run lengths were increased to 105 iterations for analyses involving a large number of simulated populations (say, when K ≥ 10) or for analyses of empirical datasets. For unbalanced sampling, the alternative ancestry model instead of the default model was used by setting POPALPHAS = 1.Both ADMIXTURE and sNMF were developed specifically for diallelic markers and could not analyse multiallelic marker data. In this study, they were used to analyse SNP data only. For the human 1000 genome phase I data, however, ADMIXTURE could not complete the analysis within a realistic period of time (72 h, the maximum allowed in the linux cluster used for the analysis) even when the maximal number of parallel threads were used. Therefore, only sNMF and PopCluster were used to analyse this dataset.To understand the relative computational efficiency and how much speedup can be gained by parallelisation, ADMIXTURE, sNMF and PopCluster were used to analyse the HGDP dataset and the 1000 genome dataset, by using a variable number of parallel threads on a linux cluster with many nodes, each having 32 cores. The maximum wall clock time allowed for a job on the cluster is 48 h. More

The world’s population is rapidly urbanizing, particularly along coastlines, where population density is now three times higher than the global average1,2. According to the National Oceanic and Atmospheric Administration (NOAA), almost 40% of the United States (U.S.) population resides in coastal zones with population density being over five times greater in coastal shoreline counties than the national average. As a result, human encroachment on coastal ecosystems is significantly modifying natural landscapes and reducing intact coastal habitat. Along densely populated coastlines, residential development often involves unsustainable land-use planning and armoring of shorelines, where natural habitats such as saltmarshes, mangroves, seagrasses and oyster reefs, are replaced with artificial structures, including vertical bulkheads, seawalls, boat ramps, and other gray infrastructures3. In areas with dense residential development between 50–90% of shorelines can be armored, whereby, on average, 14% of all U.S. shorelines have been modified from their natural conditions and replaced with artificial structures3. This transition represents an extensive loss of natural coastal habitats and the critical ecosystem services they provide.As more ecologically harmful infrastructure is developed to meet the demands of human population growth, urbanization concurrently alters ecosystem services and functions by negatively impacting biodiversity, ecological conditions and environmental quality, specifically through a decrease in native habitat, increased water pollution, and creation of impervious surfaces4. Urbanization may also lead to less resilient and adaptable coastal communities against natural hazards and climate change threats, such as sea level rise and hurricanes. This is because in urban areas, ecosystem functioning is reduced and associated services are lost, resulting in increasing risk of shoreline erosion, saltwater intrusion, storm surges, and coastal flooding2,5.These human-environment interactions in coastal ecosystems can lead to, and at the same time be derived by, decisions that will shape the future structure, function, and sustainability of coastal ecosystems6. These social decisions (e.g., large-scale policies or individual level choices) can have long-lasting consequences for both the environment and society, especially as coastal development increases. Decisions that modify and change the biophysical nature of the environment (e.g., waterfront residents’ decision to use artificial structures for storm protection and shoreline stabilization) impact its ecological functionality7. At the same time, these alterations may change the degree of connectivity that individual humans have to their environments, which might extend to broader societies’ ecological knowledge8,9.Few studies provide evidence that the removal and lack of natural environments in urbanized environments reduces individuals’ environmental connectedness and ecological knowledge, and subsequently lowers pro-environmental behaviors10,11,12. This is of critical importance since a general lack of environmental connectedness, and in particular, a lack of ecological knowledge is a phenomenon often used to explain the non-appreciation of, and deleterious behaviors toward, the natural environment, even though many studies theorize these relationships opposed to empirically test them (e.g., see refs. 13,14 and the discussion in ref. 15 about “nature-deficit disorder”). Furthermore, if there exists a general lack of ecological knowledge, social decisions at the individual level that reflect these limited perceptions (e.g., utilitarian land-use decisions12 or waterfront homeowners’ preference to install a bulkhead) can often cascade to larger societal impacts through domino-effects, where individual decisions trigger similar, reactive decisions by neighbors leading to broader societal patterns16. For example, Gittman et al.17 found that one of the stronger predictors of an individual decision to have an armored shoreline was presence of armoring on a neighboring parcel. When considered across a community scale, such societal patterns can alter natural coastal habitats significantly.In this current study, we investigate the relationship between residents’ knowledge, or mental models, of human-environment interactions, their self-reported pro-environmental behavior, and how these perceptions and behaviors are associated with urbanization. A mental model is the cognitive internal representation of a system in the external world that articulates causal relationships among system components (i.e., abstract concepts)18,19. Mental models that represent causal knowledge can be graphically obtained through cognitive mapping techniques in the form of directed graphs, which are networks in which nodes represent concepts (i.e., system components) and graph edges (arrows) represent the causal relationships between the concepts20. We combine methods from social science, data science, and network science to conduct an analysis using mental models of coastal residents along an urbanization gradient to better understand the interconnections among urbanization, people’s knowledge of human-environment interactions, and their pro-environmental behavior.We surveyed residents across eight coastal states in the northeast U.S., including Maine, New Hampshire, Massachusetts, Rhode Island, Connecticut, New York, New Jersey, and Delaware. We used a fuzzy cognitive mapping (FCM) approach21 to elicit mental models of coastal ecosystems with a focus on environmental connectedness, ecosystem health, human wellbeing, climate, and sustainable coasts (see Methods). Here, we propose a concept, Urbanized Knowledge Syndrome (UKS), which represents recurring patterns in urban dwellers’ mental models about natural ecosystems – their internal understanding of how humans and environment interact. Here, syndrome should not be interpreted as a set of medical signs and symptoms which are associated with a particular disease or disorder. These recurring patterns include (1) diminished systems thinking (e.g., complexity of mental models decreases as degree of urban development increases) and (2) the erosion of cognitive diversity (i.e., diversity of mental models among residents decreases as degree of urban development increases). These patterns demonstrate a type of thinking that is simplified to some extent or otherwise limited or focused on fewer social-ecological relationships than exist in reality.Systems thinking – a holistic view that considers factors and interactions and how they result in a possible outcome – is an important skillset that helps people better understand complex systems and adapt to changes22. Individuals with higher degrees of systems thinking are more likely to consider interdependencies, identify leverage points to intervene within the system and produce desired outcomes23, better anticipate system function and emergence of patterns of behavior24, and avoid unintended consequences18. As such, systems thinking may help coastal residents develop mental models that enable more nuanced reasoning about diverse causal pathways between humans and natural coastal ecosystems25,26,27,28, which may lead to behaviors that are driven by more predominant cognition of complex feedbacks, trade-offs, and reciprocal interdependencies between humans and nature. In contrast, bounded systems thinking (or linear thinking) may lead humans to develop limited cognition of their surrounding world, reduce their ability to accurately and adequately perceive the complexity of the environment they inhabit and interact with22, and thus may give rise to counterproductive behaviors and decisions27,29. For example, a simple causal relationship might be that seawall construction increases coastal protection as a form of structural defense to control shoreline erosion; whereas a more complex relationship might be that seawalls lead to alterations in hydrodynamic processes, which reduces erosion locally and accelerates coastal erosion downstream30, and at the same time, shoreline armoring can also lead to losses of natural coastal habitats and their critical ecosystem functions3.While cities are beneficial to human development, working as engines of socioeconomic change, cultural transformation, and technological innovation, their psychological influences on people and how these influences drive urban residents’ perceptions and behavior must be noted. Firstly, the salience of ecosystem services is limited for inhabitants of more urbanized areas, as compared to rural areas. Exposure to nature provides multiple opportunities for cognitive development which increases the potential for stewardship of the environment and for a stronger recognition of ecosystem functions13. Urban residents, however, are more routinely exposed to built environment and gray infrastructure, such as armored shorelines and artificial structures along coastlines, as opposed to natural environment, and thus their local experience of, and connection to, ecosystem services can be limited31.Secondly, urbanization generally comes with complex technology and commerce, allowing individuals to meet their needs quickly and through many choices with less appreciation of, and first hand experience with, provisioning ecosystem services (e.g., food comes from many grocery stores not a farm or garden; fish comes over a counter not across a dock or the end of a spear; and potable water comes from a pipeline not a spring or well). This may cause the development of a wider gap in human perceptions of benefits received from natural ecosystems32, fostering the emergence of societies that are increasingly disconnected and seemingly independent from ecosystem services31.Finally, residents of urbanized areas may be exposed to a set of social norms, information, and perspectives that encourage anthropocentric values and thinking including human exemptionalism (“the tendency to see human systems as exempt from the constraints of natural environment”33) and human exceptionalism (“the tendency to see humans as biologically unique and discontinuous with the rest of the animal world”34), therefore limiting their understanding of the importance and substantiality of reciprocal interdependencies between humans and natural environment13,34. These urbanization aspects may spark what we call ‘limits to systems thinking’ in the social-ecological realm.Therefore, we hypothesize (H1) that in more urbanized areas, mental models are predominantly characterized by linear thinking of coastal ecosystems, as opposed to systems thinking, where components are connected mostly by simple causal patterns. This class of mental models is associated with limited cognition of synergies and trade-offs, emergence of global patterns from local relationships, reciprocal interdependencies, and feedback loops between humans and natural ecosystems, which may lead to a gap in residents’ perception of nonlinear complex structures. To test our hypothesis, we analyze the structure of causal relationships using the network structure and graph-theoretic metrics of cognitive maps (i.e., graphical representations of mental models). We use cluster analysis to identify predominant classes of mental models about coastal ecosystems. Distinct clusters of mental models represent archetypal cognitions that individuals develop to perceive human-environment interdependencies13,27. We then use network analyses to measure the complexity of causal structures in cognitive maps and determine the overall degree of systems thinking in each cluster (see Methods). Finally, we investigate the association between urbanization and the degree of systems thinking across those clusters.The second important feature that helps systems adapt to changes is diversity, ranging from ecosystems35 to economic systems36. There is also evidence that these same relationships between diversity and adaptability hold true for cultural knowledge systems, governance systems, and among diverse communities and social institutions that function more effectively as resilient collectives28,37,38.In contrast, as cultural homogenization theories explain, survival in cities depends on fitting in and adopting practices that are considered socially normal by the dominant culture39. Although cities are magnets for people from all corners of the world with seemingly more diverse composition of race and ethnicity compared to rural areas40, assimilation of diverse values, beliefs, cultural knowledge, and social norms into a universal, governing culture—sometimes referred to as “cultural colonialism” or “cultural normalization” – is a major component of urban societies41. This cultural normalization among urban dwellers is exacerbated by dominant exposure to the universal language and education system, greater access to the Internet, social media and news outlets, and market-driven policies and global standardizations for laws and finance41.In addition, an important characteristic of urbanization is the centralization of the population into cities, “where neighborhoods in different regions have similar patterns of roads, residential lots, commercial areas, and aquatic features”42. Such physical and environmental homogenization across urban areas, which is visually evident, is influenced by monocentric land-management and policies, economic pressures for land development and use, engineering necessities, codes and standards, and preferences for particular aesthetics and recreations. Prior studies have shown that this homogenization extends to ecological structure, meaning that across urbanized areas, similar built environment and landscape structures can lead to homogenized ecological characteristics, function, and the range of ecosystem services they can supply42,43.Here, we argue that homogenization in cultural, physical, and ecological systems also extends to residents’ perceptions and understanding of human-environment interactions. We, therefore, hypothesize (H2) that increased urbanization is associated with more homogenized mental models of coastal ecosystems. To test our second hypothesis, we measure the structural dissimilarity of individuals’ mental models (i.e., cognitive maps) using some of the widely used methods for comparing graphs44. We measure the mean of pairwise cognitive distances (i.e., a quantitative metric that represents the mean of graph dissimilarity between any two individual cognitive maps) and compare this metric across clusters of mental models, and thus, explore the correspondence between urbanization and mental model homogenization (i.e., testing the hypothesis that urbanization is associated with more similar mental models in terms of causal structures represented in cognitive maps) (see Methods). More

The haplotype networks, PCoA analysis and STRUCTURE analysis based on the six nuclear loci confirm that S. longifolium is a hybrid between S. emersum and S. gramineum, providing molecular support for previous morphological analyses5. Furthermore, all individuals with intermediate admixture coefficient (Fig. 2b) and private haplotypes only present in one out of six nuclear loci (Fig. 1) suggest that S. longifolium is most likely a F1 hybrid. We thus hypothesized that S. emersum and S. gramineum could likely maintain their species boundary through the post-zygote reproductive isolation mechanism of F1 generation sterility. This hypothesis is possible based on the observations from hybrids in European Russia. The pollen viability was checked in S. longifolium samples from Vysokovskoe Lake and Sabro Lake, and the vast majority of checked pollens were sterile5. In addition, flowering plants of S. longifolium often do not form seeds, or the seeds are puny and significantly inferior to normal seeds in size5. However, the hypothesis is only based on our limited sampling, which is contrary to the conclusion inferred from morphological characteristics that it is fertile and may backcross with parental species1. Further studies with extensive sampling are necessary to test our hypothesis.The chloroplast DNA fragment trnH-psbA was used to infer the direction of hybridization between S. emersum and S. gramineum because chloroplast DNA is maternal inheritance in Sparganium3,4. The hybrid S. longifolium shared haplotypes with S. emersum and S. gramineum simultaneously (Fig. 1). This finding clearly indicates that bidirectional hybridization exists between S. emersum and S. gramineum. At the same time, the different frequency of these two haplotypes in the hybrid (H1, 19.1% vs. H2, 80.9%) means that the direction of hybridization is asymmetric. A variety of factors can lead to asymmetry in natural hybridization, such as flowering time, preference of pollinators, quality and quantity of pollen, cross incompatibility and the abundance of parent species7,8. Rare species usually act as maternal species relative to abundant species9,10. S. gramineum is confined to oligotrophic lakes and its abundance is obviously lower than that of S. emersum1,11. The relatively scarcity combined with the ecology of S. gramineum make it more often act as maternal species when hybridizing with S. emersum.As described by5, the morphological diversification of S. longifolium was also observed in this study. For example, individuals of S. longifolium with emergent and floating-leaved life forms occur concurrently in Zaozer’ye Lake (Supplementary Fig. S2). However, all individuals had the same haplotype H2 as S. gramineum (Fig. 1), suggesting that the direction of hybridization do not determine life form of S. longifolium. In addition, all individuals of S. longifolium sampled here are likely F1 hybrid. Their variable phenotypes could not be associated with traits segregation due to F2 generation or backcross. Detailed ecological investigation combining with research at the genomic level are essential to find out the potential factors leading to morphological diversification of S. longifolium.Here, using sequences of six nuclear loci and one chloroplast DNA fragment, we confirmed that S. longifolium is the hybrid between S. emersum and S. gramineum. The natural hybridization between S. emersum and S. gramineum is bidirectional but the latter mainly acts as maternal species. We also found that all samples of S. longifolium were F1 generations, indicating that S. emersum and S. gramineum could maintain their species boundary through the post-zygote reproductive isolation mechanism of F1 generation sterility. More