Ecology

Gryseels, S., Bruyn, L. D., Gyselings, R., Leendertz, H. & Leirs, H. Risk of human-to-wildlife transmission of SARS-CoV-2. Mammal Rev. 51, 272–292 (2020).Article 

Google Scholar 
Townsend, A. K., Hawley, D. M., Stephenson, J. F. & Williams, K. E. G. Emerging infectious disease and the challenges of social distancing in human and non-human animals: EIDs and sociality. Proc. R. Soc. B Biol. Sci. 287, 20201039 (2020).CAS 
Article 

Google Scholar 
Dickman, A. J. From Cheetahs to Chimpanzees: A comparative review of the drivers of human–carnivore conflict and human–primate conflict. Folia Primatol. 83, 377–387 (2013).Article 

Google Scholar 
Nyhus, P. J. Human–wildlife conflict and coexistence. Annu. Rev. Environ. Resour. 41, 143–171 (2016).Article 

Google Scholar 
Cunningham, A. A. One health, emerging infectious diseases and wildlife. Philos. Trans. R. Soc. Lond. B Biol. Sci. 372, 4 (2017).
Google Scholar 
Daszak, P., Cunningham, A. A. & Hyatt, A. D. Emerging infectious diseases of wildlife—Threats to biodiversity and human health. Science 287, 443–449 (2000).ADS 
CAS 
PubMed 
Article 

Google Scholar 
Fagre, A. C. et al. Assessing the risk of human-to-wildlife pathogen transmission for conservation and public health. Ecol. Lett. https://doi.org/10.1111/ele.14003 (2022).Article 
PubMed 

Google Scholar 
Messenger, A. M., Barnes, A. N. & Gray, G. C. Reverse zoonotic disease transmission (Zooanthroponosis): A systematic review of seldom-documented human biological threats to animals. PLoS One 9, 1–9 (2014).
Google Scholar 
Craft, M. E. Infectious disease transmission and contact networks in wildlife and livestock. Philos. Trans. R. Soc. B Biol. Sci. 370, 20140107 (2015).Article 

Google Scholar 
Bradley, C. A. & Altizer, S. Urbanization and the ecology of wildlife diseases. Trends Ecol. Evol. 22, 95–102 (2007).PubMed 
Article 

Google Scholar 
Balasubramaniam, K. N., Huffman, M. A., Sueur, C. & Macintosh, A. J. J. Primate infectious disease ecology: Insights and future directions at the human–macaque interface. In The Behavioral Ecology of the Tibetan Macaque. Fascinating Life Sciences (eds Li, J. et al.) 249–284 (Springer, 2020).Chapter 

Google Scholar 
McCabe, C. M., Reader, S. M. & Nunn, C. L. Infectious disease, behavioural flexibility and the evolution of culture in primates. Proc. R. Soc. B Biol. Sci. 282, 20140862 (2014).Article 

Google Scholar 
Silk, M. J. et al. Integrating social behaviour, demography and disease dynamics in network models: Applications to disease management in eclining wildlife populations. Philos. Trans. R. Soc. B Biol. Sci. 374, 20180211 (2019).Article 

Google Scholar 
Engel, G. A. & Jones-Engel, L. The role of Macaca fascicularis in infectious disease transmission. In Monkeys on the Edge: Ecology and Management of Long-Tailed Macaques and Their Interface with Humans (eds Gumert, M. D. et al.) 183–203 (Cambridge University Press, 2011).Chapter 

Google Scholar 
Anderson, R. M. & May, R. M. Infectious Diseases of Humans: Dynamics and Control (Oxford University Press, 1992).
Google Scholar 
Drewe, J. A. & Perkins, S. E. Disease transmission in animal social networks. In Animal Social Networks (eds Krause, J. et al.) 95–110 (Oxford University Press, 2015).
Google Scholar 
Godfrey, S. S. Networks and the ecology of parasite transmission: A framework for wildlife parasitology. Int. J. Parasitol. Parasites Wildl. 2, 235–245 (2013).PubMed 
PubMed Central 
Article 

Google Scholar 
Gomez, J. M., Nunn, C. L. & Verdu, M. Centrality in primate–parasite networks reveals the potential for the transmission of emerging infectious diseases to humans. Proc. Natl. Acad. Sci. 110, 7738–7741 (2013).ADS 
CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Godfrey, S. S., Bull, C. M., James, R. & Murray, K. Network structure and parasite transmission in a group living lizard, the gidgee skink, Egernia stokesii. Behav. Ecol. Sociobiol. 63, 1045–1056 (2009).Article 

Google Scholar 
VanderWaal, K. L., Atwill, E. R., Isbell, L. A. & McCowan, B. Linking social and pathogen transmission networks using microbial genetics in giraffe (Giraffa camelopardalis). J. Anim. Ecol. 83, 406–414 (2014).PubMed 
Article 

Google Scholar 
Drewe, J. A. Who infects whom? Social networks and tuberculosis transmission in wild meerkats. Proc. R. Soc. B Biol. Sci. 277, 633–642 (2010).Article 

Google Scholar 
MacIntosh, A. J. J. et al. Monkeys in the middle: Parasite transmission through the social network of a wild primate. PLoS One 7, 15–21 (2012).
Google Scholar 
Epstein, J. & Axtell, R. Growing Artificial Societies: Social Science from the Bottom Up (MIT Press, 1996).Book 

Google Scholar 
Bansal, S., Grenfell, B. T. & Meyers, L. A. When individual behaviour matters: Homogeneous and network models in epidemiology. J. R. Soc. Interface 4, 879–891 (2007).PubMed 
PubMed Central 
Article 

Google Scholar 
Brauer, F. Compartmental models in epidemiology, chapter 2. In Mathematical Epidemiology (eds Brauer, F. et al.) (Springer, 2008).MATH 
Chapter 

Google Scholar 
Carne, C., Semple, S., MacLarnon, A., Majolo, B. & Maréchal, L. Implications of tourist–macaque interactions for disease transmission. EcoHealth 14, 704–717 (2017).PubMed 
PubMed Central 
Article 

Google Scholar 
Rushmore, J. et al. Network-based vaccination improves prospects for disease control in wild chimpanzees. J. R. Soc. Interface 11, 20140349 (2014).PubMed 
PubMed Central 
Article 

Google Scholar 
Sah, P., Mann, J. & Bansal, S. Disease implications of animal social network structure: A synthesis across social systems. J. Anim. Ecol. 87, 546–558 (2018).PubMed 
Article 

Google Scholar 
Griffin, R. H. & Nunn, C. L. Community structure and the spread of infectious disease in primate social networks. Evol. Ecol. 26, 779–800 (2012).Article 

Google Scholar 
Hasegawa, M., Kishino, H. & Yano, T. Dating of the human–ape splitting by a molecular clock of mitochondrial DNA. J. Mol. Evol. 22, 160–174 (1985).ADS 
CAS 
PubMed 
Article 

Google Scholar 
Fuentes, A. & Hockings, K. J. The ethnoprimatological approach in primatology. Am. J. Primatol. 72, 841–847 (2010).PubMed 
Article 

Google Scholar 
Lappan, S., Malaivijitnond, S., Radhakrishna, S., Riley, E. P. & Ruppert, N. The human–primate interface in the new normal: Challenges and opportunities for primatologists in the COVID-19 era and beyond. Am. J. Primatol. 82, 1–12 (2020).Article 
CAS 

Google Scholar 
Mckinney, T. A classification system for describing anthropogenic influence on nonhuman primate populations. Am. J. Primatol. 77, 715–726 (2015).PubMed 
Article 

Google Scholar 
Devaux, C. A., Mediannikov, O., Medkour, H. & Raoult, D. Infectious disease risk across the growing human–non human primate interface: A review of the evidence. Front. Public Health 7, 1–22 (2019).Article 

Google Scholar 
Kaur, T. & Singh, J. Primate-parasitic zoonoses and anthropozoonoses: A literature review. In Primate Parasite Ecology: The Dynamics and Study of Host–Parasite Relationships (eds Huffman, M. A. & Chapman, C. A.) 199–230 (Cambridge University Press, 2009).
Google Scholar 
Melin, A. D., Janiak, M. C., Marrone, F., Arora, P. S. & Higham, J. P. Comparative ACE2 variation and primate COVID-19 risk. Commun. Biol. 3, 641 (2020).CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Klegarth, A. Synanthropy. In The International Encyclopedia of Primatology (Wiley, 2017). https://doi.org/10.1002/9781119179313.wbprim0448.Chapter 

Google Scholar 
Gumert, M. D. A common monkey of Southeast Asia: Longtailed macaque populations, ethnophoresy, and their occurrence in human environments. In Monkeys on the Edge: Ecology and Management of Longtailed Macaques and Their Interface with Humans (eds Gumert, M. D. et al.) 3–43 (Cambridge University Press, 2011).Chapter 

Google Scholar 
Riley, E. P. The human–macaque interface: Conservation implications of current and future overlap and conflict in Lore Lindu National Park, Sulawesi, Indonesia. Am. Anthropol. 109, 473–484 (2007).Article 

Google Scholar 
Thierry, B. Unity in diversity: Lessons from macaque societies. Evol. Anthropol. 16, 224–238 (2007).Article 

Google Scholar 
Balasubramaniam, K. N. et al. The influence of phylogeny, social style, and sociodemographic factors on macaque social network structure. Am. J. Primatol. 80, e227227 (2018).Article 

Google Scholar 
Sueur, C. et al. A comparative network analysis of social style in macaques. Anim. Behav. 82(4), 845–852 (2011).Article 

Google Scholar 
Balasubramaniam, K. N. et al. Implementing social network analysis to understand the socioecology of wildlife co-occurrence and joint interactions with humans in anthropogenic environments. J. Anim. Ecol. 90, 2819–2833 (2021).PubMed 
Article 

Google Scholar 
Henzi, S. P. & Barrett, L. The value of grooming to female primates. Primates 40, 47–59 (1999).Article 

Google Scholar 
Schino, G. & Aureli, F. Trade-offs in primate grooming reciprocation: Testing behavioural flexibility and correlated evolution. Biol. J. Linn. Soc. 95, 439–446 (2008).Article 

Google Scholar 
Radhakrishna, S. & Sinha, A. Less than wild? Commensal primates and wildlife conservation. J. Biosci. 36, 749–753 (2011).PubMed 
Article 

Google Scholar 
Balasubramaniam, K. N. et al. Impact of individual demographic and social factors on human–wildlife interactions: A comparative study of three macaque species. Sci. Rep. 10, 21991 (2020).ADS 
CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Marty, P. R. et al. Time constraints imposed by anthropogenic environments alter social behaviour in long-tailed macaques. Anim. Behav. 150, 157–165 (2019).Article 

Google Scholar 
Kaburu, S. S. K. et al. Interactions with humans impose time constraints on urban-dwelling rhesus macaques (Macaca mulatta). Behaviour 156, 1255–1282 (2019).Article 

Google Scholar 
Altmann, J. Observational study of behavior: Sampling methods. Behaviour 49, 227–267 (1974).CAS 
PubMed 
Article 

Google Scholar 
Kaburu, S. S. K. et al. Rates of human–monkey interactions affect grooming behaviour among urban-dwelling rhesus macaques (Macaca mulatta). Am. J. Phys. Anthropol. 168, 92–103 (2019).PubMed 
Article 

Google Scholar 
Martin, P. & Bateson, P. Measuring Behaviour (Cambridge University Press, 1993).Book 

Google Scholar 
Farine, D. R. & Whitehead, H. Constructing, conducting and interpreting animal social networks. J. Anim. Ecol. 84, 1144–1163 (2015).PubMed 
PubMed Central 
Article 

Google Scholar 
Rozins, C. et al. Social structure contains epidemics and regulates individual roles in disease transmission in a group-living mammal. Ecol. Evol. 8, 12044–12055 (2018).PubMed 
PubMed Central 
Article 

Google Scholar 
Fujii, K., Jin, J., Shev, A., Beisner, B., McCowan, B. & Fushing, H. Perc: Using percolation and conductance to find information flow certainty in a direct network (R Package Version 0.1.2.) https://rdrr.io/cran/Perc/ (2016).Funkhouser, J. A., Mayhew, J. A., Sheeran, L. K. & Mulcahy, J. B. comparative investigations of social context-dependent dominance in captive chimpanzees (Pan troglodytes) and wild Tibetan macaques (Macaca thibetana). Sci. Rep. 8, 1–15 (2018).CAS 
Article 

Google Scholar 
McCowan, B. J. et al. Measuring dominance certainty and assessing its impact on individual and societal health in a nonhuman primate: A network approach. Philos. Trans. R. Soc. B 377, 20200438 (2022).Article 

Google Scholar 
Bjornstad, O. N. Package ‘epimdr’ (2020).Tuite, A. R. et al. Estimated epidemiologic parameters and morbidity associated with pandemic H1N1 influenza. CMAJ 182, 131–136 (2010).PubMed 
PubMed Central 
Article 

Google Scholar 
Arienzo, M. D. & Coniglio, A. Assessment of the SARS-CoV-2 basic reproduction number, R0, based on the early phase of COVID-19 outbreak in Italy. Biosaf. Health 2, 57–59 (2020).PubMed 
PubMed Central 
Article 

Google Scholar 
Bailey, N. T. The Mathematical Theory of Epidemics (Griffin, 1957).
Google Scholar 
Magnusson, A., Skaug, H., Nielsen, A., Berg, C., Kristensen, K., Maechler, M., van Bentham, K., Sadat, N., Bolker, B. & Brooks, M. Package ‘glmmTMB’. https://cran.r-project.org/web/packages/glmmTMB/glmmTMB.pdf (2019).Quinn, G. P. & Keough, M. J. Experimental Designs and Data Analysis for Biologists (Cambridge University Press, 2002).Book 

Google Scholar 
Lüdecke, D., Ben-Shachar, M., Patil, I., Waggoner, P. & Makowski, D. Performance: An R package for assessment, comparison and testing of statistical models. J. Open Source Softw. 6, 3139 (2021).ADS 
Article 

Google Scholar 
Chiyo, P. I., Moss, C. J. & Alberts, S. C. The influence of life history milestones and association networks on crop-raiding behavior in male African elephants. PLoS One 7, e31382 (2012).ADS 
CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
VanderWaal, K. L., Atwill, E. R., Isbell, L. A. & McCowan, B. Quantifying microbe transmission networks for wild and domestic ungulates in Kenya. Biol. Conserv. 169, 136–146 (2014).Article 

Google Scholar 
Berman, C. M. Primate kinship: Contributions from Cayo Santiago. Am. J. Primatol. 78, 63–77 (2016).PubMed 
Article 

Google Scholar 
Balasubramaniam, K. N. et al. Social network community structure and the contact-mediated sharing of commensal E. coli among captive rhesus macaques (Macaca mulatta). PeerJ 6, e4271 (2018).PubMed 
PubMed Central 
Article 

Google Scholar 
Marty, P. R. et al. Individuals in urban dwelling primate species face unequal benefits associated with living in an anthropogenic environment. Primates 61, 245–259 (2020).Article 

Google Scholar 
Zinsstag, J., Schelling, E., Waltner-Toews, D. & Tanner, M. From ‘one medicine’ to ‘one health’ and systemic approaches to health and well-being. Prev. Vet. Med. 101, 148–156 (2011).CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Lloyd-Smith, J. O., Schreiber, S. J., Kopp, P. E. & Getz, W. M. Superspreading and the effect of individual variation on disease emergence. Nature 438, 355–359 (2005).ADS 
CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Schülke, O. et al. Quantifying within-group variation in sociality—covariation among metrics and patterns across primate groups and species. Behav. Ecol. Sociobiol. 76, 50 (2022).Article 

Google Scholar 
Romano, V., Shen, M., Pansanel, J., MacIntosh, A. J. J. & Sueur, C. Social transmission in networks: Global efficiency peaks with intermediate levels of modularity. Behav. Ecol. Sociobiol. 72, 154 (2018).Article 

Google Scholar  More

Bolnick, D. I. et al. The ecology of individuals: Incidence and implications of individual specialization. Am. Nat. 161, 1–28. https://doi.org/10.1086/343878 (2003).MathSciNet 
Article 
PubMed 

Google Scholar 
Peterson, A. T., Soberón, J. & Sánchez-Cordero, V. Conservatism of ecological niches in evolutionary time. Science 285, 1265–1267 (1999).CAS 
Article 

Google Scholar 
Wiens, J. J. & Graham, C. H. Niche conservatism: Integrating evolution, ecology, and conservation biology. Annu. Rev. Ecol. Evol. Syst. 36, 519–539 (2005).Article 

Google Scholar 
Bolnick, D. I. et al. Why intraspecific trait variation matters in community ecology. Trends Ecol. Evol. 26, 183–192. https://doi.org/10.1016/j.tree.2011.01.009 (2011).Article 
PubMed 
PubMed Central 

Google Scholar 
Gentile, G., Bonelli, S. & Riva, F. Evaluating intraspecific variation in insect trait analysis. Ecol. Entomol. 46, 11–18 (2021).Article 

Google Scholar 
Ortego, J., Calabuig, G., Cordero, P. J. & Aparicio, J. M. Egg production and individual genetic diversity in lesser kestrels. Mol. Ecol. 16, 2383–2392 (2007).CAS 
Article 

Google Scholar 
Peacor, S. D., Schiesari, L. & Werner, E. E. Mechanisms of nonlethal predator effect on cohort size variation: Ecological and evolutionary implications. Ecology 88, 1536–1547 (2007).Article 

Google Scholar 
Smith, A. B., Godsoe, W., Rodríguez-Sánchez, F., Wang, H.-H. & Warren, D. Niche estimation above and below the species level. Trends Ecol. Evol. 34, 260–273 (2019).Article 

Google Scholar 
Carlson, B. S., Rotics, S., Nathan, R., Wikelski, M. & Jetz, W. Individual environmental niches in mobile organisms. Nat. Commun. 12, 4572. https://doi.org/10.1038/s41467-021-24826-x (2021).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Hutchinson, G. E. Population studies: Animal ecology and demography. Cold Spring Harb. Symp. Quant. Biol. 22, 415–427 (1957).Article 

Google Scholar 
Venter, O. et al. Global terrestrial Human Footprint maps for 1993 and 2009. Sci. Data 3, 160067 (2016).Article 

Google Scholar 
Watson, J. E. M., Dudley, N., Segan, D. B. & Hockings, M. The performance and potential of protected areas. Nature 515, 67–73 (2014).ADS 
CAS 
Article 

Google Scholar 
Wauchope, H. S. et al. Protected areas have a mixed impact on waterbirds, but management helps. Nature 605, 103 (2022).CAS 
Article 

Google Scholar 
Lowry, H., Lill, A. & Wong, B. B. Behavioural responses of wildlife to urban environments. Biol. Rev. 88, 537–549 (2013).Article 

Google Scholar 
Hällfors, M. H. et al. Combining range and phenology shifts offers a winning strategy for boreal Lepidoptera. Ecol. Lett. 24, 1619–1632 (2021).Article 

Google Scholar 
Joppa, L. N. & Pfaff, A. Global protected area impacts. Proc. R. Soc. B Biol. Sci. 278, 1633–1638. https://doi.org/10.1098/rspb.2010.1713 (2011).Article 

Google Scholar 
Hanson, J. O. et al. Global conservation of species’ niches. Nature 580, 232–234 (2020).ADS 
CAS 
Article 

Google Scholar 
Geldmann, J., Manica, A., Burgess, N. D., Coad, L. & Balmford, A. A global-level assessment of the effectiveness of protected areas at resisting anthropogenic pressures. Proc. Natl. Acad. Sci. 116, 23209–23215. https://doi.org/10.1073/pnas.1908221116 (2019).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Blonder, B. Hypervolume concepts in niche- and trait-based ecology. Ecography 41, 1441–1455 (2018).Article 

Google Scholar 
Mammola, S. & Cardoso, P. Functional diversity metrics using kernel density n-dimensional hypervolumes. Methods Ecol. Evol. 11, 986–995. https://doi.org/10.1111/2041-210X.13424 (2020).Article 

Google Scholar 
Mammola, S. Assessing similarity of n-dimensional hypervolumes: Which metric to use? J. Biogeogr. 46, 2012 (2019).Article 

Google Scholar 
Carvalho, J. C. & Cardoso, P. Decomposing the causes for niche differentiation between species using hypervolumes. Front. Ecol. Evol. https://doi.org/10.3389/fevo.2020.00243 (2020).Article 

Google Scholar 
Pavlek, M. & Mammola, S. Niche-based processes explaining the distributions of closely related subterranean spiders. J. Biogeogr. 48, 118–133. https://doi.org/10.1111/jbi.13987 (2021).Article 

Google Scholar 
Wang, X. et al. Exploring ecological specialization in pipefish using genomic, morphometric and ecological evidence. Divers. Distrib. 27, 1393–1406. https://doi.org/10.1111/ddi.13286 (2021).Article 

Google Scholar 
Jaturapruek, R., Fontaneto, D., Mammola, S. & Maiphae, S. Potential niche displacement in species of aquatic bdelloid rotifers between temperate and tropical areas. Hydrobiologia. https://doi.org/10.1007/s10750-021-04681-z (2021).Article 

Google Scholar 
Hu, Z. M. et al. Intraspecific genetic variation matters when predicting seagrass distribution under climate change. Mol. Ecol. 30, 3840–3855. https://doi.org/10.1111/mec.15996 (2021).Article 
PubMed 

Google Scholar 
Ho, D. E., Imai, K., King, G. & Stuart, E. A. Matching as nonparametric preprocessing for reducing model dependence in parametric causal inference. Polit. Anal. 15, 199–236 (2007).Article 

Google Scholar 
Terraube, J., Van Doninck, J., Helle, P. & Cabeza, M. Assessing the effectiveness of a national protected area network for carnivore conservation. Nat. Commun. 11, 2957. https://doi.org/10.1038/s41467-020-16792-7 (2020).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Chichorro, F., Juslén, A. & Cardoso, P. A review of the relation between species traits and extinction risk. Biol. Conserv. 237, 220–229 (2019).Article 

Google Scholar 
Pearman, P. B., Guisan, A., Broennimann, O. & Randin, C. F. Niche dynamics in space and time. Trends Ecol. Evol. 23, 149–158 (2008).Article 

Google Scholar 
Santangeli, A., Högmander, J. & Laaksonen, T. Returning white-tailed eagles breed as successfully in landscapes under intensive forestry regimes as in protected areas. Anim. Conserv. 16, 500–508. https://doi.org/10.1111/acv.12017 (2013).Article 

Google Scholar 
Broennimann, O. et al. Evidence of climatic niche shift during biological invasion. Ecol. Lett. 10, 701–709 (2007).CAS 
Article 

Google Scholar 
Fitzpatrick, M. C., Weltzin, J. F., Sanders, N. J. & Dunn, R. R. The biogeography of prediction error: Why does the introduced range of the fire ant over-predict its native range? Glob. Ecol. Biogeogr. 16, 24–33 (2007).Article 

Google Scholar 
Dietz, H. & Edwards, P. J. Recognition that causal processes change during plant invasion helps explain conflicts in evidence. Ecology 87, 1359–1367 (2006).Article 

Google Scholar 
Holt, R. D., Keitt, T. H., Lewis, M. A., Maurer, B. A. & Taper, M. L. Theoretical models of species’ borders: Single species approaches. Oikos 108, 18–27 (2005).Article 

Google Scholar 
Zhang, Z., Mammola, S., McLay, C. L., Capinha, C. & Yokota, M. To invade or not to invade? Exploring the niche-based processes underlying the failure of a biological invasion using the invasive Chinese mitten crab. Sci. Total Environ. 728, 138815. https://doi.org/10.1016/j.scitotenv.2020.138815 (2020).ADS 
CAS 
Article 
PubMed 

Google Scholar 
Liu, C., Wolter, C., Xian, W. & Jeschke, J. M. Most invasive species largely conserve their climatic niche. Proc. Natl. Acad. Sci. 117, 23643–23651 (2020).CAS 
Article 

Google Scholar 
Sarasola, J. H., Grande, J. M. & Negro, J. J. Birds of Prey: Biology and Conservation in the XXI Century 63–94 (Springer, 2018).Book 

Google Scholar 
Reif, J., Hořák, D., Krištín, A., Kopsová, L. & Devictor, V. Linking habitat specialization with species’ traits in European birds. Oikos 125, 405–413. https://doi.org/10.1111/oik.02276 (2016).Article 

Google Scholar 
Thornton, D., Branch, L. & Sunquist, M. Passive sampling effects and landscape location alter associations between species traits and response to fragmentation. Ecol. Appl. 21, 817–829. https://doi.org/10.1890/10-0549.1 (2011).Article 
PubMed 

Google Scholar 
Hatfield, J. H., Orme, C. D. L., Tobias, J. A. & Banks-Leite, C. Trait-based indicators of bird species sensitivity to habitat loss are effective within but not across data sets. Ecol. Appl. 28, 28–34. https://doi.org/10.1002/eap.1646 (2018).Article 
PubMed 

Google Scholar 
Kuuluvainen, T. Forest management and biodiversity conservation based on natural ecosystem dynamics in Northern Europe: The complexity challenge. Ambio 38, 309–315 (2009).Article 

Google Scholar 
Niemi, J. & Ahlstedt, J. Finnish Agriculture and Rural Industries 2011 (MTT Economic Research, Agrifood Research Finland, 2011).
Google Scholar 
Lehikoinen, P. et al. Increasing protected area coverage mitigates climate-driven community changes. Biol. Cons. 253, 108892. https://doi.org/10.1016/j.biocon.2020.108892 (2021).Article 

Google Scholar 
Virkkala, R. & Lehikoinen, A. Patterns of climate-induced density shifts of species: Poleward shifts faster in northern boreal birds than in southern birds. Glob. Change Biol. 20, 2995–3003. https://doi.org/10.1111/gcb.12573 (2014).ADS 
Article 

Google Scholar 
Lehikoinen, A. & Virkkala, R. North by north-west: Climate change and directions of density shifts in birds. Glob. Change Biol. 22, 1121–1129. https://doi.org/10.1111/gcb.13150 (2016).ADS 
Article 

Google Scholar 
Santangeli, A., Rajasärkkä, A. & Lehikoinen, A. Effects of high latitude protected areas on bird communities under rapid climate change. Glob. Change Biol. 23, 2241–2249. https://doi.org/10.1111/gcb.13518 (2017).ADS 
Article 

Google Scholar 
Lehikoinen, P., Santangeli, A., Jaatinen, K., Rajasärkkä, A. & Lehikoinen, A. Protected areas act as a buffer against detrimental effects of climate change—Evidence from large-scale, long-term abundance data. Glob. Change Biol. 25, 304–313. https://doi.org/10.1111/gcb.14461 (2019).ADS 
Article 

Google Scholar 
Santangeli, A. & Lehikoinen, A. Are winter and breeding bird communities able to track rapid climate change? Lessons from the high North. Divers. Distrib. 23, 308–316. https://doi.org/10.1111/ddi.12529 (2017).Article 

Google Scholar 
Lindén, H., Helle, E., Helle, P. & Wikman, M. Wildlife triangle scheme in Finland: Methods and aims for monitoring wildlife populations. Finnish Game Res. 49, 4–11 (1996).
Google Scholar 
Blonder, B. Do hypervolumes have holes? Am. Nat. 187, E93–E105. https://doi.org/10.1086/685444 (2016).Article 
PubMed 

Google Scholar 
Fuller, C., Ondei, S., Brook, B. W. & Buettel, J. C. First, do no harm: A systematic review of deforestation spillovers from protected areas. Glob. Ecol. Conserv. 18, e00591. https://doi.org/10.1016/j.gecco.2019.e00591 (2019).Article 

Google Scholar 
Hyvärinen, E., Juslén, A., Kemppainen, E., Uddström, A. & Liukko, U.-M. Suomen lajien uhanalaisuus–Punainen kirja 2019 (2019).Wilman, H. et al. EltonTraits 1.0: Species-level foraging attributes of the world’s birds and mammals. Ecology 95, 2027–2027. https://doi.org/10.1890/13-1917.1 (2014).Article 

Google Scholar 
Morelli, F., Benedetti, Y., Møller, A. P. & Fuller, R. A. Measuring avian specialization. Ecol. Evol. 9, 8378–8386 (2019).Article 

Google Scholar 
Jetz, W., Thomas, G. H., Joy, J. B., Hartmann, K. & Mooers, A. O. The global diversity of birds in space and time. Nature 491, 444 (2012).ADS 
CAS 
Article 

Google Scholar 
Cimatti, M. et al. Large carnivore expansion in Europe is associated with human population density and land cover changes. Divers. Distrib. 27, 602–617. https://doi.org/10.1111/ddi.13219 (2021).Article 

Google Scholar 
Laaksonen, T. & Lehikoinen, A. Population trends in boreal birds: Continuing declines in agricultural, northern, and long-distance migrant species. Biol. Conserv. 168, 99–107. https://doi.org/10.1016/j.biocon.2013.09.007 (2013).Article 

Google Scholar 
Blonder, B., Lamanna, C., Violle, C. & Enquist, B. J. The n-dimensional hypervolume. Glob. Ecol. Biogeogr. 23, 595–609. https://doi.org/10.1111/geb.12146 (2014).Article 

Google Scholar 
Cardoso, P. M., Rigal, F. & Carvalho, J. BAT-Biodiversity Assessment Tools (2014).Zuur, A. F. & Ieno, E. N. A protocol for conducting and presenting results of regression-type analyses. Methods Ecol. Evol. 7, 636–645 (2016).Article 

Google Scholar 
Brooks, M. E. et al. glmmTMB balances speed and flexibility among packages for zero-inflated generalized linear mixed modeling. R J. 9, 378–400 (2017).Article 

Google Scholar 
Zuur, A. F., Ieno, E. N. & Elphick, C. S. A protocol for data exploration to avoid common statistical problems. Methods Ecol. Evol. 1, 3–14. https://doi.org/10.1111/j.2041-210X.2009.00001.x (2010).Article 

Google Scholar 
Sokal, R. R., Rohlf, F. J. & Rohlf, J. F. Biometry (Macmillan, 1995).MATH 

Google Scholar 
Dormann, C. F. et al. Collinearity: A review of methods to deal with it and a simulation study evaluating their performance. Ecography 36, 27–46. https://doi.org/10.1111/j.1600-0587.2012.07348.x (2013).Article 

Google Scholar 
Lüdecke, D., Ben-Shachar, M. S., Patil, I., Waggoner, P. & Makowski, D. performance: An R package for assessment, comparison and testing of statistical models. J. Open Source Softw. https://doi.org/10.21105/joss.03139 (2021).Article 

Google Scholar 
Revell, L. J. phytools: An R package for phylogenetic comparative biology (and other things). Methods Ecol. Evol. 3, 217–223 (2012).Article 

Google Scholar 
Zuur, A., Ieno, E. N., Walker, N., Saveliev, A. A. & Smith, G. M. Mixed Effects Models and Extensions in Ecology with R 1–552 (Springer, 2009).Book 

Google Scholar 
R Core Development Team. R: A Language and Environment for Statistical Computing. (R Foundation for Statistical Computing, 2021). https://www.R-project.org/. More

Renewal processes represent development under variable conditionsThe consequence of a drastic environmental change can be demonstrated by introducing a shift in development time during the process. For demonstration, we consider a scenario where a group of individuals enter into a favourable environment reducing development time from (40pm 5) time units to (20pm 5).We show, in Fig. 1, that our dynamic pseudo-stage-structured MPM yields a gradual stage completion with an average development time of approximately (30pm 5) steps (solid dark lines) when conditions shift at ({tau }=20) (each step corresponds to 1 time unit). The target Erlang-distributed development trajectories without the shift are shown as dashed gray lines. The snapshots of the population structure, represented by the development indicator q, taken at each time step, show that half of the development is complete at the time of the switch and the switch accelerates the accumulation of q (Fig. S1).Figure 1Response to change in development time. The number of developing individuals is simulated by using the cumulative development process and compared to (a) the age-dependent development process, (b) an ODE representation, (c) an LCT representation, and (d) a DDE representation. Solid dark lines show the cumulative development and thick blue lines show the alternative models. Dashed gray lines mark the two target trajectories before and after the shift in development time (marked with red crosses).Full size imageIn age-dependent development, a sharp transition, instead of a gradual one, is observed at the (20^{th}) step (Fig. 1a). The switch results in the majority of individuals reaching target development age immediately at the time of switch. Previous work, reported in Erguler et al.59 and Erguler et al.55, aimed at modelling population dynamics under variable conditions, based on this dynamic age-dependent framework. Our results suggest that cumulative development might improve the fit to the data, prediction accuracy, and applicable geospatial range of these models.We see in Fig. 1b that the canonical ODE framework represents an exponentially distributed development time and a shift in rate at (t=20). The LCT extension to the framework helps to incorporate time dependence and represent the long and short development time distributions (Fig. 1c). The resulting model accommodates change in the rate parameter (gamma ) (Eq. 8), e.g doubling of (gamma ) changes development time from (40pm 5) to (20pm 2.5). However, to accommodate the required shift, the model needs to be transformed from a 66-dimensional system to an 18-dimensional one, which is beyond the scope of this work. We argue that in cases where development time distribution is fixed a priori (excluded from model calibration), the LCT framework provides a significant advantage over canonical ODEs. Although the framework has been used in the field of infectious disease epidemiology64,65, it has recently been applied to the modelling of vector population dynamics30.The DDE framework also yields a gradual development trajectory with an intermediate duration (Fig. 1d). However, the distribution tends towards the longer development trajectory compared to the one achieved with cumulative development. The canonical DDE framework assumes a homogenous cohort, where all individuals react in the same way to variations in development rate. The assumption gives rise to sharp stage transitions within a single generation if all individuals are introduced at the same time. As a potential workaround, it has been proposed to generate a plausible population history, through variable entry times, until the required (or observed) developmental variation builds up31,32. Variation in development rates then acts upon the population and results in the modification of the existing age-structure. It is worthwhile to mention that a recent extension to the DDE framework to accommodate trait variation in population dynamics34 might also accommodate changing development rates within a single stage; however, it has not yet been employed at this scale.Cumulative development is in agreement with the widely known degree-day (DD) framework, where development time is predicted by the heat accumulating in organisms46. Although the rate of accumulation in response to environmental conditions varies considerably, the DD framework implies that the combination of two different rates yields an average development time (also seen with cumulative development in Fig. 1). Experimental evaluation of this will be the topic of future research.It is worth mentioning that our dynamically structured renewal process-based MPM follows the assumption of random population heterogeneity9,11; namely, at the individual level, the future behaviour of an organism is not affected by its historical behaviour. However, trait variation within a population is prevalent in many species, and is known to impact population dynamics and species interactions34,66,67. Future development of our framework will consider improving upon this limitation.Environmental variation transformed into development timesSeveral non-linear relationships have been proposed to represent the temperature dependence of insect development68. A common feature is the presence of low and high temperature thresholds beyond which development is prohibitively slow. Often, there exists an optimum between the thresholds where the process is most efficient. A typical relationship between temperature and development rate, reported in Briere et al.50, is seen in Fig. 2a. Mean development time, given by the reciprocal of rate in Fig. 2b, exhibits the two thresholds and the optimum.Figure 2Development under environmental variation. In (a), development rate (Eq. 9) is shown with (alpha =1.5times 10^{-5}), (T_L=0^oC), and (T_H=50^oC). In (b), mean development time is shown together with the probability densities of three temperature regimes ((rho _L), (rho _M), and (rho _H)). In (c), the number of individuals completing development at each step are shown with respect to the three temperature regimes. Solid lines indicate the median, shaded areas indicate the (90%) range of 1000 simulations, and thick lines indicate simulations with the expected values of each regime.Full size imageTo investigate how temperature variation is transformed into cumulative development time, we assumed three variation regimes at relatively low, medium, and high temperatures ((rho _L), (rho _M), and (rho _H), respectively). Densities of the corresponding Gaussian probability distributions are plotted in Fig. 2b. Accordingly, each variation is transformed by a slightly different region of the rate function (Eq. 9). Eventually, the three development time distributions emerge as shown in Fig. 2c.We found that the output of (rho _H) is skewed towards longer durations compared to what we would otherwise obtain if we simulated the process under constant conditions with the mean of (rho _H). The impact of variation in the middle range, (rho _M), is similar to that of (rho _H), but less pronounced. Conversely, the output of (rho _L) is skewed towards shorter durations. Our results suggest that, when development is already highly efficient, variation in temperature causes frequent encounters of longer (but not shorter) development durations, eventually extending the overall duration of the process. In the low efficiency range, development takes long to complete, but frequent encounters of relatively short durations—especially as the process approaches its optimum duration—triggers completion earlier than in the case of no variation.Overall, our model predictions are in agreement with the rate summation effect, which states that the different outcomes obtained under constant and varying temperatures is due to the non-linear relationship between temperature and development rate (the Kaufmann effect)16. Furthermore, acceleration of development in insects subjected to varying high temperatures, its retardation at varying low temperatures, and low variability of development time in the linear range of the rate curve have been widely discussed69. Several groups have reported evidence in support of this effect, which is also in agreement with our results. For instance, Vangansbeke et al. (2015) reported for three insect species, Phytoseiulus persimilis, Neoseiulus californicus, and Tetranychus urticae, that varying temperatures with a lower mean yields faster development compared to the yield at mean constant temperatures70. However, observations of this phenomenon might result in different responses for different species at similar temperatures due to the difference in rate curves. Identification of the optimum temperature range may facilitate comparison. For instance, Carrington et al. (2013) assumed (26^oC) as optimum based on the high dengue incidence in Thailand, and showed that large variations around (26^oC) increases development time for the dengue vector, Aedes aegypti71. Wu et al. (2015) demonstrated that development is faster at around (26^oC) compared to (23^oC) for the fly, Megaselia scalaris, and found that varying temperatures at around (23^oC) accelerates the process47. Finally, in a modelling study employing DDs, Chen et al. (2013) reported that larger diurnal temperature ranges relate to additional DD accumulation and faster development in grape berry moth, Paralobesia viteana72. Under the realistic non-optimum field conditions, where these simulations had been performed, a decrease in development time is expected in response to varying temperatures according to our results.We note that the variation in development times is due to temperature since we ignore intrinsic stochasticity to demonstrate the impact of (rho ) in isolation. The deterministic setup removes the upper limit in the number of distinct pseudo-stage indicators: a different q emerges from each k, and a different k emerges from each (rho ). Since the number of pseudo-stages quickly exhausts the computational resources, we set the precision of q to the nearest 100(^{th}) decimal point, effectively capping the number of pseudo-stages at 100 (see Accuracy of the pseudo-stage approximation). As shown in Fig. S2, the approximation has a negligible impact on accuracy.Environmental dependency extracted from life tables under constant conditionsHaving discussed the importance of environmental variability in development, in this section, we employ a well-established experimental method to unravel the relationship between temperature and development time in a common mosquito species. In contrast to invasive vectors, which effectively render new territories suitable for disease transmission, Culex species pose an imminent threat with their wide distribution and ornitophilic (Cx. pipiens biotype pipiens), mamophilic (Cx. pipiens biotype molestus), and intermixed (their hybrids) blood feeding behaviour. Here, we investigate the temperature dependencies of mortality and development of Cx. quinquefasciatus, the southern house mosquito, which is an important disease vector, widely distributed across the tropics and sub-tropics73,74.To infer the dependencies, we used a generic temperature-driven insect development model, described in Methods, and the life history observations performed at five constant temperatures (15, 20, 23, 27, and (30,^{circ })C) under laboratory conditions60,61. As a result of the inverse modelling procedure, detailed in Methods, we found that the generic model yields an overall match between the simulations and observations. In Fig. 3a, we present a comparison of observed and simulated maximum production and the stage-emergence times for pupae and adults. Here, we define the stage-emergence time as the time taken from the beginning of an experiment to the time when half of the maximum production of a stage (pupa or adult) is observed. In addition, in Fig. S3, we present the comparison of time trajectories separately for each temperature.Figure 3Inverse modelling of Cx. quinquefasciatus environmental dependency. The comparison of observed and simulated maximum pupa (P) and adult (A) production and the corresponding stage-emergence times is given in (a). Observations are represented with dots and simulations with box plots. The environmental dependency of larva and pupa development time (b) and mortality (c), derived by the posterior mode sample (Theta _q), is shown in (b,c). Solid lines represent the median and shaded areas represent the (90%) range.Full size imageWe found that the generic model faithfully replicates the observed development times of larvae and pupae. On the other hand, stage mortalities are predicted well at three temperatures, but are overestimated at 20 or (27,^{circ })C. The impact of temperature on mortality might be more complex than it is captured by the quartic equation (Eq. 11). Optimum survival seen at (27,^{circ })C suggests that the relationship might be non-symmetrical or multimodal. In addition, the observed variability in mortality suggests that the mismatch could also be due to experimental error or the intrinsic stochasticity of the biological processes.We extracted the functional forms of temperature dependence from the posterior samples, shown in Fig. 3b, c, and found that the data inform the model as expected within the temperature range of the experiments ((15{-}30,^{circ })C). Stage durations are well informed, and reflect the low variability seen in the data (the standard deviation is less than 1.5 days at all temperatures for both stages). Accordingly, pupae develop in less than 4 days, which is much shorter than the larva development time (between 10 and 20 days above (20,^{circ })C). The model predicts that the minimum temperature at which development occurs (from the larva stage) is (10.5,^{circ })C, which is close to (10.9,^{circ })C, reported in Grech et al.75.The observed variability in pupa and adult production suggests that survival is a highly stochastic process regardless of the controlled laboratory conditions. A deterministic model, such as the one used in this context, represents the mean of such processes but does not capture their variability. The simulated variability is a result of the uncertainty in parameter estimates. Model parameters contribute unequally to the output as a result of the model structure and the functional forms of temperature dependence, and the data inform certain parameters better than others76,77. For instance, daily mortality, shown in Fig. 3c, is more constrained for larva than pupa, which is likely due to the short duration of the pupa stage—changes in daily mortality have larger consequences as development time increases.We note that a well-informed model yields predictions in the form of verifiable hypotheses; however, these are not necessarily accurate predictions. Model accuracy is assessed when such hypotheses are experimentally tested as part of the cyclic process of model development78. Here, we demonstrated that our modelling framework can be used to derive biologically meaningful inferences and to help improve the understanding of the temperature dependence of Cx. quinquefasciatus.Greater information content of semi-field experimentsThe number of experiments required to test a range of conditions, including different combinations of multiple drivers, may quickly exhaust available resources. Moreover, variable conditions may have a previously unaccounted impact on development and mortality. In this section, we demonstrate that observations performed under variable conditions are valuable sources of information for our modelling framework, which is capable of representing the dynamics under such conditions.Cx. pipiens, the northern house mosquito, is a competent disease vector, widely distributed across the temperate countries in North America, Europe, Asia, and North and East Africa74,79. Unlike Cx. quinquefasciatus, Cx. pipiens biotype pipiens is known to enter a reproductive diapause phase, where adult females arrest oogenesis during harsh winter conditions80,81. When larvae are exposed to short photoperiods and low temperatures during development, they emerge as adults destined to diapause. Although Cx. pipiens biotype molestus has lost the ability to diapause, its immature stages have been reported to retain metabolic sensitivity to photoperiod82,83.To reveal the environmental dependence of the molestus biotype, we exposed its eggs to variable temperatures in semi-field conditions until adult emergence (or loss of cohort). The numbers of viable larvae, pupae, and adults observed in different experimental batches are given in Fig. S4. We employed the extended model with both temperature and photoperiod dependence (see Methods), and calibrated the model against seven of the semi-field experiments, performed in March, May, June, July, August, and September (Fig. S4(a), (b), (d), (f), (g), (i) and (j)).As a result, we found that the model replicates the patterns of abundance emerging in the observations, e.g. stage timing and maximum adult production, reasonably well in most of the experiments, regardless of the times during which they were performed (Figs. S5 and S6). Quantitative evaluation of the agreement reveals that the observed and simulated adult emergence times are less than a week apart (Table 1).Table 1 Comparison of observed and simulated adult emergence time and the total number of adults produced. Simulation output is given in terms of the median and (90%) range.Full size tableOn the other hand, we found that egg and larva mortalities, and also, pupa and adult production are highly variable in the observations (see Fig. S4(c), (f), and (g)). Spikes of larva mortality are seen in Spring and Autumn (especially in May, September, and October). Despite this variability, the difference between the predicted and observed adult production was around 11 or less, except in the case of the experiment E7, which unexpectedly yielded only one pupa and no adults.We obtain relatively large mismatches when predicting larva abundances, specifically where egg mortality is not predicted well (E5, E7, E8, E10, E11, E12). We hypothesise that the stress associated with rearing lab-grown specimens under variable conditions might elevate egg mortality, induce premature hatching, or affect the survival of the larvae produced. Since egg development starts inside gravid females, i.e. under the optimum conditions of the laboratory, the observable part of development subjected to variable conditions remains mainly the hatching behaviour. Consequently, we observed rapid and synchronous completion of the egg stage in all experiments (see Figs. S5 and S6). Being exposed to a narrow range of temperatures, relatively less information can be obtained on the environmental dependency of the egg stage. As a potential improvement, we recommend that future adaptations of the semi-field experiments consider using field-captured adult female mosquitoes as the source of eggs.In addition to egg mortality, we observed spikes of larva mortality in May (E3), July (E8), and in Autumn (E14, E15, and E16). A likely cause of such transient high mortality is brief temperature shifts towards the extremes. However, the rarity of such events prevents the inverse modelling procedure from adequately capturing their impacts on life processes. As a potential improvement, we recommend that the experiments are performed in overlapping time frames, increasing the likelihood of observing the impact of an extreme event at different times during development. We note that the early decline in larva abundance seen in Autumn could be a result of insufficient food supply due to the increased nutritional requirements. According to the proposed metabolic response to short photoperiods, larvae would require additional food to accumulate fat reserves in preparation for diapause, the state where adult females endure several months without feeding. This implies that development takes longer than it would at long photoperiods when subjected to similar temperature regimes.Using the extended model and the semi-field data, we identified the environmental dependencies shown in Fig. 4. The data informed about the temperature dependency of each life stage as well as the photoperiod dependency of larvae. As expected, the overall variability in the inferred dependencies is higher for Cx. pipiens compared to Cx. quinquefasciatus (Fig. 3). We found that the larva and pupa development times closely match the observations reported by Spanoudis et al.62 at long photoperiods (see Fig. S7). However, the development times reported in Kiarie-Makara et al.84 at short photoperiods and moderate temperatures do not suggest a significant impact of daylight, which could be due to the particular strain of Cx. pipiens used in these experiments. As expected, the temperature dependency of egg development was not well informed by the data in the current configuration of the model and the functional forms of environmental dependence.Figure 4Environmental dependency of Cx. pipiens development and mortality inferred from semi-field life table experiments. Solid lines represent the median and shaded areas represent the (90%) range.Full size imageWe found that the photoperiod dependency is significantly non-linear with an average threshold of 13.7 hours of daylight (Fig. 4c). Photoperiod-driven extension in development time (about 1.7 times more at 13:11 h L:D than at 15:9 h L:D) contributes to improving the accuracy of predictions at the end of the high season (Fig. S8). The critical photoperiod (CPP) agrees well with the ones identified for Cx. pipiens biotype pipiens85,86. For instance, Sanburg and Larsen reported that there is an exponential relationship between follicle sizes in adult females (signifying commitment to diapause) and the photoperiods they were exposed to during immature stages85. We inferred a similar (but reverse) gradient between photoperiod and the extension of larva development time from 15 to 12 hours of daylight (Fig. 4c).Risk assessment with annual development profilesWe extrapolated the development dynamics of Cx. pipiens over the calendar year by setting up a hypothetical experiment at the beginning of each week. We simulated the subsequent development dynamics and obtained the annual development profile as shown in Fig. 5. Accordingly, the immature stages begin development from late February and the first adults emerge in May (adults emerging late in May start developing in the experiments set up late in March). The profile is consistent with the regular Cx. pipiens high season in the region.Figure 5Annual development profile of Cx. pipiens in Petrovaradin, Serbia, in 2017. The outcome of each hypothetical semi-field experiment is plotted vertically along the y-axis at the date when the experiment is initiated. The maximum number of adults produced is given in blue, and the time it takes (from the date indicated on the x-axis) to produce half of the maximum is given in green. Solid lines represent the median and shaded areas represent the 90% range of model predictions. Outcomes of the semi-field experiments (dots) are plotted together with the model predictions. The time points marked with circles indicate the experiments used to calibrate the model. Estimated time of first adult emergence is given in the inset.Full size imageAs seen in Fig. 5, predicted adult emergence times agree well with the observations throughout the high season. However, there is a greater variation in the maximum number of adults than the times of emergence (extending to almost (40%) of the possible outcomes in early August). A greater variability (almost (80%) in August) is seen in the corresponding observations, which we transformed into the percentage of eggs emerging as adults (where available) to facilitate comparison. According to the model, variation in adult production is associated with the variation in both development times and mortality during immature stages. We recall that the uncertainty in the informed environmental dependencies is high around relatively less frequently encountered values—especially the lower and higher temperature extremes (Fig. 4). Specifically, egg development times cannot be identified precisely, but immediate hatching of the larvae is predicted between 20 and 25 °C. Consequently, we found that frequent exposure to temperatures outside the well-informed range have a significant impact on the variation in adult production (Fig. S9).We adopt the time of first adult emergence as a proxy of the first generation of adults in the season. According to our model, early adult emergence is a result of shorter development times and higher success rates, which indicates that the temperature conditions allow for an early first generation of adults. An early first generation greatly contributes to an early peak of adult abundance, which may increase the risk of vector-borne disease transmission in humans. For instance, an early peak of abundance may cause an early start of West Nile virus circulation and amplification in Culex pipiens and their avian hosts, which increases the likelihood of virus spillover to humans51,87. Anecdotal evidence shows that the anomalously hot April and May that occurred in 2018 in Serbia shifted the peak of Cx. pipiens abundance forward by more than one month (Petrić et al., unpublished). Similarly, 2018 was the year with the largest number of autochthonous West Nile virus infections throughout Europe (more than the total of the previous seven years together)88,89.In summary, our results showed that the semi-field experiments, when used in combination with our dynamic pseudo-stage-structured MPM, help to develop predictive models and inform over a wide range of environmental conditions. We developed a predictive model of Cx. pipiens biotype molestus development and gained insights into the specifics of temperature and photoperiod dependencies by reducing the need of extensive laboratory data. We used life history observations from 7 experiments performed under semi-field conditions and employed a generic model structure, largely uninformed on the specific environmental dependencies of the species. The cumulative development framework we introduced applies broadly to poikilotherms subjected to highly variable environmental conditions. Although the generic model structure helps to develop exploratory models and identify potential environmental dependencies, accuracy can be improved by customising the models for the known dependencies of particular species. With a straightforward extension of the development model to cover the complete life cycle (with egg laying and density dependence), it is possible to incorporate field observations of eggs or adult mosquitoes, and develop an environment-driven population dynamics model. More

Ma, Z., Xue, L. & Du, S. Theory and Technology of High-Efficient Use of Water and Fertilizer for Watermelon and Melon in Gravel-Mulched Field 59–62 (Science Press, Beijing, 2018).
Google Scholar 
Qiu, Y., Xie, Z., Wang, Y., Malhi, S. S. & Ren, J. Long-term effects of gravel—Sand mulch on soil organic carbon and nitrogen in the Loess Plateau of northwestern China. J. Arid. Land 7, 46–53. https://doi.org/10.1007/s40333-014-0076-7 (2015).Article 

Google Scholar 
Zhang, K., Zhang, W., Tan, L., An, Z. & Zhang, H. Effects of gravel mulch on aeolian transport: A field wind tunnel simulation. J. Arid. Land 7, 296–303. https://doi.org/10.1007/s40333-015-0121-1 (2015).Article 

Google Scholar 
Yamanaka, T., Inoue, M. & Kaihotsu, I. Effects of gravel mulch on water vapor transfer above and below the soil surface. Agric. Water Manag. 67, 145–155. https://doi.org/10.1016/j.agwat.2004.01.002 (2004).Article 

Google Scholar 
Wang, J., Xie, Z., Guo, Z. & Wang, Y. Simulating the effect of gravel-sand mulched field degradation on soil temperature and evaporation. J. Desert Res. 30, 6 (2010).
Google Scholar 
Kaseke, K. F. et al. The effects of desert pavements (gravel mulch) on soil micro-hydrology. Pure Appl. Geophys. 169, 873–880. https://doi.org/10.1007/s00024-011-0367-2 (2012).ADS 
Article 

Google Scholar 
Inagaki, M. N. How does a stone mulch increase transpiration and grain yield in wheat under soil water deficit stress?. Cereal Res. Commun. 40, 486–493 (2012).CAS 
Article 

Google Scholar 
Abdelfattah, M. A. Pedogenesis, land management and soil classification in hyper-arid environments: Results and implications from a case study in the United Arab Emirates. Soil Use Manag. 29, 279–294 (2013).Article 

Google Scholar 
Lightfoot, D. The cultural ecology of Puebloan pebble-mulch gardens. Hum. Ecol. 21, 115–143. https://doi.org/10.1007/BF00889356 (1993).Article 

Google Scholar 
Graf, A., Kuttler, W. & Werner, J. Mulching as a means of exploiting dew for arid agriculture?. Atmos. Res. 87, 369–376 (2008).Article 

Google Scholar 
Shao Ping, D. U., Ma, Z. M. & Xue, L. Distribution characteristics of soil aggregates and their associated organic carbon in gravel-mulched land with different cultivation years. Ying Yong Sheng Tai Xue Bao 28, 1619–1625 (2017).
Google Scholar 
Zhong-Ming, M. A., Shao-Ping, D. U. & Xue, L. Influences of sand-mulching years on soil temperature, water content, and growth and water use efficiency of watermelon. J. Desert Res. 33, 1433–1439 (2013).
Google Scholar 
Pang, L. et al. Effect of different gravel mulched years on soil microflora and physicochemical properties in gravel-sand mulched field. Agric. Res. Arid Areas (2017).Hao, H. et al. Effects of gravel-sand mulching on soil bacterial community and metabolic capability in the semi-arid Loess Plateau, China. World J. Microbiol. Biotechnol. 33, 209 (2017).PubMed 
Article 
CAS 

Google Scholar 
Gregorich, E. G., Carter, M. R., Angers, D. A., Monreal, C. & Ellert, B. H. Towards a minimum data set to assess soil organic matter quality in agricultural soils. Can. J. Soil Sci. 74, 367–385 (1994).CAS 
Article 

Google Scholar 
Jandl, R. et al. Current status, uncertainty and future needs in soil organic carbon monitoring. Sci. Total Environ. 468–469, 376–383 (2014).ADS 
PubMed 
Article 
CAS 

Google Scholar 
Lal, R. Enhancing crop yields in the developing countries through restoration of the soil organic carbon pool in agricultural lands. Land Degrad. Dev. 17, 197–209 (2010).Article 

Google Scholar 
Razafimbelo, T. M. et al. Aggregate associated-C and physical protection in a tropical clayey soil under Malagasy conventional and no-tillage systems. Soil Tillage Res. 98(2), 140–149 (2007).Article 

Google Scholar 
Hongbing, Z. et al. Effect of long-term tillage on soil aggregates and aggregate-associated carbon in black soil of Northeast China. PLoS ONE 13, e0199523 (2018).Article 
CAS 

Google Scholar 
Bajracharya, R. M., Lal, R. & Kimble, J. M. Soil organic carbon distribution in aggregates and primary particle fractions as influenced by erosion phases and landscape position. In Soil Processes & the Carbon Cycle (eds Lal, R. et al.) (CRC Press, 1998).
Google Scholar 
Sekaran, U., Sagar, K. L. & Kumar, S. Soil aggregates, aggregate-associated carbon and nitrogen, and water retention as influenced by short and long-term no-till systems. Soil Tillage Res. 208, 104885 (2020).Article 

Google Scholar 
Tang, X., Liu, S., Liu, J. & Zhou, G. Effects of vegetation restoration and slope positions on soil aggregation and soil carbon accumulation on heavily eroded tropical land of Southern China. J. Soils Sediments 10, 505–513 (2010).CAS 
Article 

Google Scholar 
Wang, Y. et al. 23-Year manure and fertilizer application increases soil organic carbon sequestration of a rice–barley cropping system. Biol. Fertil. Soils 51(5), 583–591 (2015).Article 

Google Scholar 
Zhou, H., Fang, H., Mooney, S. J. & Peng, X. Effects of long-term inorganic and organic fertilizations on the soil micro and macro structures of rice paddies. Geoderma 266, 66–74 (2016).ADS 
CAS 
Article 

Google Scholar 
Congreves, K. A., Hooker, D. C., Hayes, A., Verhallen, E. A. & Eerd, L. V. Interaction of long-term nitrogen fertilizer application, crop rotation, and tillage system on soil carbon and nitrogen dynamics. Plant Soil 410, 113–127 (2017).CAS 
Article 

Google Scholar 
Tang, H., Xiao, X., Chao, L., Ke, W. & Pan, X. Impact of long-term fertilization practices on the soil aggregation and humic substances under double-cropped rice fields. Environ. Sci. Pollut. Res. Int. 25, 11034–11044 (2018).CAS 
PubMed 
Article 

Google Scholar 
Rong, Y., Su, Y. Z., Wang, T. & Qin, Y. Effect of chemical and organic fertilization on soil carbon and nitrogen accumulation in a newly cultivated farmland. J. Integr. Agric. 15, 658–666 (2019).
Google Scholar 
Chen, Y. et al. Rotation and organic fertilizers stabilize soil water-stable aggregates and their associated carbon and nitrogen in flue-cured tobacco production. J. Soil Sci. Plant Nutr. 20, 192–205 (2020).Article 
CAS 

Google Scholar 
Li, T. et al. Contrasting impacts of manure and inorganic fertilizer applications for nine years on soil organic carbon and its labile fractions in bulk soil and soil aggregates. CATENA 194, 104739. https://doi.org/10.1016/j.catena.2020.104739 (2020).CAS 
Article 

Google Scholar 
Hati, K. M. et al. 50 Years of continuous no-tillage, stubble retention and nitrogen fertilization enhanced macro-aggregate formation and stabilisation in a Vertisol. Soil Tillage Res. 214, 105163. https://doi.org/10.1016/j.still.2021.105163 (2021).Article 

Google Scholar 
Ma, P. et al. Macroaggregation is promoted more effectively by organic than inorganic fertilizers in farmland ecosystems of China—A meta-analysis. Soil Tillage Res. 221, 105394. https://doi.org/10.1016/j.still.2022.105394 (2022).Article 

Google Scholar 
Lu, R. Chemical Analysis Methods of Soil and Agriculture (China Agricultural Science and Technology Press, 2000).
Google Scholar 
Dorodnikov, M., Blagodatskaya, E., Blagodatsky, S., Marhan, S. & Kuzyakov, Y. Stimulation of microbial extracellular enzyme activities by elevated CO2 depends on soil aggregate size. Glob. Change Biol. 15, 1603–1614 (2010).ADS 
Article 

Google Scholar 
Kemper, W. D. & Rosenau, R. C. Aggregate stability and size distribution. In Methods of Soil Analysis: Part 1 Physical and Mineralogical Methods. Agronomy Monograph No. 9, ASA and SSSA 2nd edn (ed. Klute, A.) 425–442 (Wiley, 1986).
Google Scholar 
Jones, J. B. Laboratory Guide for Conducting Soil Tests and Plant Analysis (Nurse Educ, 2001).Book 

Google Scholar 
Blair, G., Lefroy, R. & Lisle, L. Soil carbon fractions based on their degree of oxidation, and the development of a carbon management index for agricultural systems. Aust. J. Agric. Res. 46, 393–406 (1995).Article 

Google Scholar 
Li, W., Zheng, Z., Li, T. & Liu, M. Distribution characteristics of soil aggregates and its organic carbon in different tea plantation age. Acta Ecol. Sin. 34, 6326–6336 (2014).
Google Scholar 
Matos, E. S., Freese, D., Böhm, C., Quinkenstein, A. & Hüttl, R. Organic matter dynamics in reclaimed lignite mine soils under Robinia pseudoacacia L. plantations of different ages in Germany. Commun. Soil Sci. Plant Anal. 43, 745–755 (2012).CAS 
Article 

Google Scholar 
Zádorová, T., Jakšík, O., Kodešová, R. & Penížek, V. Influence of terrain attributes and soil properties on soil aggregate stability. Soil Water Res. 6, 111–119 (2011).Article 

Google Scholar 
Sajjadi, S. A. & Mahmoodabadi, M. Aggregate breakdown and surface seal development influenced by rain intensity, slope gradient and soil particle size. Solid Earth 6(1), 311–321 (2015).ADS 
Article 

Google Scholar 
Yang, W. et al. Mechanical properties and soil stability affected by fertilizer treatments for an Ultisol in subtropical China. Plant Soil 363(1), 157–174 (2013).CAS 
Article 

Google Scholar 
Lal, R. Soil health and carbon management. Food Energy Secur. 5, 212–222 (2016).Article 

Google Scholar 
Jagadamma, S., Lal, R., Hoeft, R. G., Nafziger, E. D. & Adee, E. A. Nitrogen fertilization and cropping systems effects on soil organic carbon and total nitrogen pools under chisel-plow tillage in Illinois. Soil Tillage Res. 95, 348–356 (2007).Article 

Google Scholar 
Yang, X. M. et al. Long-term effects of fertilization on soil organic carbon changes in continuous corn of northeast China: RothC model simulations. Environ. Manage. 32, 459–465 (2003).CAS 
PubMed 
Article 

Google Scholar 
Russell, A. E., Laird, D. A., Parkin, T. B. & Mallarino, A. P. Impact of nitrogen fertilization and cropping system on carbon sequestration in Midwestern mollisols. Soil Sci. Soc. Am. J. 69, 413–422 (2005).ADS 
CAS 
Article 

Google Scholar 
Hartwig, N. L. Cover crop and living mulches. Weed Sci. 50, 688–699 (2002).CAS 
Article 

Google Scholar 
Yu, H. et al. Effects of long-term compost and fertilizer application on stability of aggregate-associated organic carbon in an intensively cultivated sandy loam soil. Biol. Fertil. Soils 48, 325–336 (2012).CAS 
Article 

Google Scholar 
Li, C., Yan, L. & Tang, L. The effects of long-term fertilization on the accumulation of organic carbon in the deep soil profile of an oasis farmland. Plant Soil 369, 645–656 (2013).CAS 
Article 

Google Scholar 
Puget, P., Chenu, C. & Balesdent, J. Dynamics of soil organic matter associated with particle-size fractions of water-stable aggregates. Eur. J. Soil Sci. 51, 595–605 (2000).Article 

Google Scholar 
Ashman, M. R., Hallett, P. D. & Brookes, P. C. Are the links between soil aggregate size class, soil organic matter and respiration rate artefacts of the fractionation procedure?. Soil Biol. Biochem. 35, 435–444 (2003).CAS 
Article 

Google Scholar 
Six, J., Elliott, E. T. & Paustian, K. Soil macroaggregate turnover and microaggregate formation: a mechanism for C sequestration under no-tillage agriculture. Soil Biol. Biochem. 32, 2099–2103 (2000).CAS 
Article 

Google Scholar 
Chivenge, P. P., Murwira, H. K., Giller, K. E., Mapfumo, P. & Six, J. Long-term impact of reduced tillage and residue management on soil carbon stabilization: Implications for conservation agriculture on contrasting soils. Soil Tillage Res. 94, 328–337 (2006).Article 

Google Scholar 
Kölbl, A. & Knabner, I. K. Content and composition of free and occluded particulate organic matter in a differently textured arable Cambisol as revealed by solid-state 13C NMR spectroscopy. J. Plant Nutr. Soil Sci. 167, 45–53 (2004).Article 
CAS 

Google Scholar 
Tong, L. et al. Response of organic carbon fractions and microbial community composition of soil aggregates to long-term fertilizations in an intensive greenhouse system. J. Soils Sediments 20, 641–652. https://doi.org/10.1007/s11368-019-02436-x (2020).CAS 
Article 

Google Scholar 
Li, C., Li, Y. & Tang, L. The effects of long-term fertilization on the accumulation of organic carbon in the deep soil profile of an oasis farmland. Plant Soil 369, 645–656. https://doi.org/10.1007/s11104-013-1605-4 (2013).CAS 
Article 

Google Scholar 
Su, Y. Z., Wang, F., Suo, D. R., Zhang, Z. H. & Du, M. W. Long-term effect of fertilizer and manure application on soil-carbon sequestration and soil fertility under the wheat–wheat–maize cropping system in northwest China. Nutr. Cycl. Agroecosyst. 75, 285–295 (2006).CAS 
Article 

Google Scholar 
Du, S. P., Ma, Z. M. & Xue, L. Effect of manure combined with chemical fertilizers on fruit yield, fruit quality and water and nitrogen use efficiency in watermelon grown in gravel-mulched field. J. Fruit Sci. 37, 10. https://doi.org/10.13925/j.cnki.gsxb.20190380 (2020).CAS 
Article 

Google Scholar 
Zhengchao, Z., Zhuoting, G., Zhouping, S. & Fuping, Z. Effects of long-term repeated mineral and organic fertilizer applications on soil organic carbon and total nitrogen in a semi-arid cropland. Eur. J. Agron. 45, 20–26. https://doi.org/10.1016/j.eja.2012.11.002 (2013).CAS 
Article 

Google Scholar 
Lv, W. et al. Effects of organic fertilizers on continuous cropping watermelon growth and soil microflora. Acta Agric. Shanghai 22, 96–98 (2006).
Google Scholar 
Zhong, W. et al. The effects of mineral fertilizer and organic manure on soil microbial community and diversity. Plant Soil 326, 511–522 (2010).CAS 
Article 

Google Scholar  More

Historical changes and current status of the Dongzhaigang mangrove areaBased on the literature and remote sensing data, we calculated the changes in the area of mangrove forests in Dongzhaigang since the 1950s presented in Fig. 2. In the last 60 years, the area of mangrove forests in Dongzhaigang has experienced large fluctuations mainly due to human destruction and protection activities such as mariculture reclamation, cofferdams, and restoration: it decreased from 3416 hm2 in 195617 to 3213 hm2 in 195919,29 and then decreased sharply to 1733 hm2 in 1983 and to 1537 hm2 in 198720,30. Since the establishment of the national nature reserve in 1986, the decline in area of Dongzhaigang mangrove has stopped19, which are now protected and restored owing to the law and regulations that prohibit human activities from destroying the mangrove resource. In 1988, the area was restored to 1809 hm2, and since the 1990s, it has no longer decreased, remaining constant at approximately 1711 hm2 (in the range of 1575–1812 hm2) based on the literature)18,20,31,32,33,34 (Fig. 2). The area of the Dongzhaigang mangrove forest in 2019 was estimated to be 1842 hm2 based on the latest 2 m resolution remote sensing data21. Hence, we wonder how SLR has impacted Dongzhaigang mangrove in the past decades. However, it is very difficult to analyze how SLR has historically impacted the spatial changes in the Dongzhaigang mangrove; the same can be said regarding the influence of human activities, such as destruction before mid-1980s and protection after 1990s. However, the dynamic changes among low plant edges in the intertidal zone can be used to analyze the impact of natural driving forces such as SLR35, based on the latest remote sensing data for the period of 1986–2020. Thus, we analyzed the dynamic changes in low mangrove edges (hereafter, the edges), which are mainly impacted by natural impact drivers, as shown in Fig. 3. The dynamic low mangrove edges represented by 1986, 2000, and 2020 reveal the changes in spatial distribution of Dongzhaigang mangrove. As shown in Fig. 3. Most of the edges along the coast of Dongzhaigang between 1986 and 2020 migrated landward, but not significantly. However, if we look at the changes in detail, some edges such as those in Daoxue, Sanjiang (purple circles in Figs. 3a,b–d,e–g) more clearly retreated landward compared to other places. Besides, some edges of Luodou along the northeastern coast of Dongzhaigang outside the reserve and an unnamed small island (pruple circles in Fig. 3a,h–j) also migrated landward very distinctly. On the contrary, the two smaller shore lines (black circles) in the northern part of Yangfeng and Daxue districts showed seaward expansion (Fig. 3a).Figure 2Changes in the mangrove area in Dongzhaigang from 1956 to 2019. The equation in the upper-right-hand corner of the plot refers to the fitting equation of historical changes in the total area of Dongzhaigang mangrove.Full size imageFigure 3The dynamic changes in low mangrove edges in Dongzhaigang from 1986 to 2020. Maps generated in ArcMap v10.0 (https://www.esri.com/en-us/home).Full size imageVertical rate of sediment accretion in mangrove wetlandsThe vertical rate of sediment accumulation in mangrove wetlands can reflect whether the mangroves can adjust the soil surface elevation change through sediment trapping to adapt to SLR6,11. The vertical sediment accretion rates at two sites of Dongzhaigang mangrove (i.e., Linshi and Daoxue villages in Fig. 1b) can be obtained from historical documents, which are 0.41 cm year−1 at LS and 0.64 cm year−1 at DX, respectively27,28. Since historical data may not be enough to reflect the vertical sediment accretion rates in time and space, we conducted a supplementary investigation on the sediment accumulation rates at site HG in Yanfeng and SJ site in Sanjiang farms, respectively (Fig. 1b), based on the assumption that they can reflect the sediment supplies from main reivers such as Yanfeng West River and Yanzhou River, respectively. Sediment accretion rates measured using 210Pbex specific activity in the cores from sites HG and SJ showed that 210Pbex decayed exponentially with increasing depth, and the R2 values of both cores were approximately 0.80 after curve fitting. This analysis resulted in vertical sediment accretion rates of 0.53 and 0.40 cm year−1 at HG and SJ, respectively (Fig. 4). Therefore, the locations of sediment cores at sites LS, DX, HG, and SJ can basically represent the whole Dongzhaigang mangrove forest area.Figure 4210Pbex activity profiles in selected cores such as from (a) station HG and (b) station SJ.Full size imageRate of relative sea level rise in Dongzhaigang mangroveThe global mean sea level (GMSL) is accelerating due to global warming-induced thermal expansion of the oceans and melting of land-based glaciers and ice caps into the sea36. Between 1901 and 2010, the GMSL rose by 0.19 m9. Coastal China is among the regions that experience the highest levels of SLR23. The rate of RSLR along China’s coast from 1980 to 2019 was 3.4 mm year−1, higher than the global average23. In the future, under the premise of increasing anthropogenic GHG emissions, global sea levels will rise rapidly, and it is projected that the GMSL may rise by 0.84 m (0.61–1.10 m) relative to the current levels by the end of the twenty-first century9. Based on the observations from the tide gauge stations in the Haikou area and model data from the Coupled Model Intercomparison Projection 5 (CMIP5), the rate of RSLR around Dongzhaigang reached 4.6 mm year−1 from 1980 to 2018. This rate is much higher than the global and China’s average values23,25 and will likely accelerate further in the future. Based on the results of the CMIP5 model simulations under different GHG emission scenarios24, the RSLR in coastal Haikou waters, including in Dongzhaigang, is expected to be significant by 2030, 2050, and 2100 for the low, intermediate, and very high GHG emission scenarios RCPs 2.6, 4.5, and 8.5, respectively (Table 1, Fig. 5). Under RCPs 2.6, 4.5, and 8.5, the sea level will rise by 65 (42–90, likely range), 75 (51–102, likely range), and 96 (70–125, likely range) cm by 2100, respectively, with the average RSLR rates of 6.84 (4.42–9.47, likely range), 7.89 (5.37–10.74, likely range), and 10.1 (7.37–13.12, likely range) mm year−1, respectively.Table 1 Estimated coastal relative sea level rise (cm) and its rate (mm year−1) in the Haikou area under different GHG emission scenarios (data from Kopp et al.24).Full size tableFigure 5Historical and future relative sea level changes along coastal Dongzhaigang, Haikou City from 1980 to 2100; the 5–95% uncertainty ranges are shaded for RCPs 2.6, 4.5, and 8.5, respectively.Full size imageImpact of relative sea level rise on Dongzhaigang mangroveMangroves cannot easily adapt to rising sea levels if the rate of GMSL rise exceeds 6.1 mm year−1 ( > 90% probability, very likely), whereas the survival threshold for mangroves is extremely likely to be exceeded ( > 95% probability, extremely likely) when the rate of GMSL exceeds 7.6 mm year−17. Although these values are based on global levels7, they still reflect the threat of SLR to local mangroves. In view of this, we further analyzed the potential impact and risks to Dongzhaigang mangrove from future SLR under different climate scenarios.Based on the predicted future rates of SLR under RCPs 2.6, 4.5, and 8.5 and on the vertical sediment accretion rates of Dongzhaigang mangrove wetlands, the mangroves are likely to be affected by rising sea levels by 2030, 2050, and 2100, respectively (Table 2, Fig. 6). Under the low GHG emission scenario (RCP 2.6), the area of the Dongzhaigang mangrove forest will only experience a small reduction: 16.40% (1.20–16.95%, likely range), 302 hm2 (22–312 hm2, likely range); 16.73% (1.20–17.82%, likely range), 308 hm2 (22–328 hm2, likely range); and 17.60% (1.14–31.02%, likely range), 324 hm2 (21–571 hm2, likely range) by 2030, 2050, and 2100, respectively (Table 2, Fig. 6a). This is because the vertical sediment accretion rate of Dongzhaigang mangrove will remain largely constant with increasing RSLR rate. Moreover, it should be noted that compared with 2030, the increase areas of mangroves inundation caused by SLR will be small by 2050 under three RCPs scenarios (Table 2). In contrast, under the intermediate and very high GHG emission scenarios (RCPs 4.5 and 8.5), Dongzhaigang mangrove is expected to be more significantly affected by SLR. Under RCP 4.5, 26.56% (16.19–40.74%, likely range) or 489 hm2 (298–750 hm2, likely range) of mangrove forest will likely be lost by the end of the century (Table 3, Fig. 6b). Under RCP 8.5, it is projected that 31.99% (18.14–50.73%, likely range) or 589 hm2 (334–934 hm2, likely range) of mangrove forest will be lost by 2100 (Table 2, Fig. 6c). Therefore, under RCPs 4.5 and 8.5, the impact of SLR on mangrove wetlands by 2100 is much higher than that of RCP 2.6, and is likely to result in  > 26% of mangroves being lost, whereas under RCP 2.6, only 17% of mangroves are likely to be lost.Table 2 Area (hm2) and percentage of future mangrove loss in Dongzhaigang under different climate scenarios (RCPs 2.6, 4.5, and 8.5) (likely ranges).Full size tableFigure 6Potential loss of mangrove forests in Dongzhaigang under different climate scenarios (RCPs 2.6, 4.5, and 8.5). Maps generated in ArcMap v10.0 (https://www.esri.com/en-us/home).Full size imageTable 3 Core stations and depths.Full size tableUnder RCP 2.6, the rate of RSLR around Dongzhaigang will reach 0.72 cm year−1 in 2030 and then decrease in 2050 and 2080 to 0.69 and 0.68 cm year−1, respectively (Table 1). However, under RCP 4.5 (8.5), by 2030, 2050, and 2100, the rate of RSLR will reach 0.72 (0.72), 0.73 (0.80), and 0.79 (10.1) cm year−1, respectively. By 2100, some mangroves in the northern part of Tashi village, the eastern part of Yanfeng, the northern part of Daoxue Village, and the northeastern part of the Sanjiang farm will likely be lost owing to SLR, and other coastal wetlands will also be impacted. Since the rate of RSLR around Dongzhaigang is higher than the global average survival threshold for mangroves (i.e., the SLR rate exceeds 7.0 mm year−1), the Dongzhaigang mangrove will be significantly affected by SLR, with a potential loss of 31–32%; however, the survival threshold will not increase (Table 2, Fig. 6). More

Site selectionOur study focused on the mixed-conifer zone of the Sierra Nevada which is dominated by Ponderosa pine (Pinus ponderosa), Jeffrey pine (Pinus jeffreyi), Incense cedar (Calocedrus decurrens), White fir (Abies concolor) and California black oak (Quercus kelloggii). Common shrubs include Deer brush (Caenothus integerrimus), Mountain whitethorn (Caenothus cordulatus), Greenleaf manzanita (Arctostaphylos patula) and Prostrate ceanothus (Caenothus prostrates). Study sites in the Eldorado National Forest near Placerville, CA (38°45′N 120°20′W), were in and near the area burned during the King Fire (Fig. 2).We sampled sites in the mixed-conifer zone between 4000–6000 ft in three burn categories: unburned, low-to-moderate severity and high severity. We selected sites that occurred in similar pre-burn habitat (moderate to dense conifer forest) based on remotely-sensed vegetation class data from the California Wildlife Habitat Relationships program (CWHR)16. No site experienced wildfire or controlled burns in the preceding century23. Burn categories were based on remote sensing relative differenced Normalized Burn Ratio (RdNBR) canopy cover calibrated for the Sierra Nevada24. We validated these burn categories as meaningful and discrete with remote sensing (immediately post fire) and field data (3 years post-fire). Monitoring Trends in Burn Severity (MTBS) maps classify burn severity with Landsat reflectance imagery of pre-fire and post-fire conditions at 30-m resolution25. MTBS assigns pixels a value based on burn severity: 0 = outside the burn boundary; 1 = unburned-low severity within the fire perimeter; 2 = low severity; 3 = moderate severity; 4 = high severity). At each site, we determined the mean MTBS value of pixels in the small-mammal trapping grid (90 × 90 m) and a 225 m buffer (the largest home range diameter for small mammals we captured)26 extended on all sides. Immediately after the fire, the three burn categories differed significantly in tree cover and remote sensing scores of burn severity25 and are still different in tree surveys three years later.We sampled 27 sites, each 4 ha in size. To minimize site-specific influences, we paired sites across treatments to account for elevation, slope, pre-burn vegetation characteristics, pre-burn forest management, ownership, and soil type. Each burn category received nine sites. We blocked sites across burn severities (one site in each burn category per block). Site locations were chosen for accessibility, but all sites were at least 50 m from access roads and at least 200 m apart (sites were >1 km apart on average). We also excluded areas that experienced or were scheduled to experience salvage logging post fire. Occasionally, field conditions at a site location did not align with remotely sensed data classifications, a site was not large enough for homogeneous sampling plots or was dominated by slopes >30 degrees. In these cases, sites were moved to nearby locations that satisfied the site selection criteria.Sampling designIn this study we report the body size, abundance and biomass density of plants and free-living animals using a variety of methods in three different fire severities. Each site consisted of a 200 m × 200 m plot (4 ha) around which sampling methods were organized (Fig. 3). We collected data for organisms with 19 different sampling methods that were scaled to the abundance and body size of targeted organisms. Some methods were not performed at all sites, and some methods were pooled across sites (within treatments) due to low sampling success. Below, we explain each sampling method and detail any variation in its application across sites. To minimize seasonal effects, we sampled all sites in a block at the same time, over 4–5 continuous days, between late June and early September 2017. Weather was consistently hot and dry during this period with no pronounced variation. All animal survey procedures were approved by UC Davis IACUC (protocol number CA-17-451736) and carried out under CDFW permit #SC-3638.Species inclusionTo evaluate the effects of wildfire-burn severity on community structure we assembled a list of organisms and life stages encountered during our sampling of the Eldorado National Forest. Every animal in our list was broken into ontogenetic life stages. Plants were broken into constituent parts (e.g. leaves, roots, seeds) rather than stages. Every organism in our lists had at least one life stage observed by this study in the Eldorado National Forest. Not all life stages in the list were directly observed, many were inferred. These life stages were suspected to occur in the habitat but were not observed (or quantified) because (1) it was impractical (e.g. mycorrhizae), (2) our sampling methods did not capture them (e.g. larval insects in plant tissues) or (3) they could not be identified to species (e.g. arthropod eggs). Stages (not species) were excluded when they did not occur in the terrestrial habitat (e.g. aquatic larvae). Stages (or parts) were lumped when they could not be distinguished in terms of their resources or consumers (e.g. parasitoid wasp eggs, larvae and pupae). Unobserved stages were omitted from a burn category community if they were feeding (e.g. not eggs) but did not have any resources. All life stages, observed or not, received a body size estimate.This approach has several benefits. First, it fills life-cycle gaps without artificially inflating species richness (e.g. having a node labelled “bird eggs”). Second, unobserved life stages may not be major biomass contributors, but they do make important contributions to food-web structure and population dynamics. Including life stages without abundance information is useful because it allows their inclusion in analyses of network structure, informs consumer-resource body-size ratios and allows comparisons to other food datasets organized around body size, but lacking abundance estimates.A comprehensive food web required the inclusion of non-living nodes like detritus which is an important resource for many consumers. Fire creates strong spatial structure in woody detritus, so we collected mass-density information on it and partitioned detritus into types (e.g., woody vs. leaf) and size classes. We did not collect mass-density information on carcasses but treated them similarly by partitioning them into logarithmic size classes to reflect their availability to different consumers (e.g., tick carcasses vs. deer carcasses). We have also included the biotic products honeydew and dung due to their importance to certain consumer groups in the system. We have not done so here but future efforts may wish to include nutrients as resources for plants to capture gradients and competition. Finally, analyses may wish to augment our lists by explicitly including fire as a consumer/herbivore as opposed to an external force shaping treatment effects.Below we describe the methods used to quantify the richness, abundance, and biomass density of these organisms in each of our burn categories. Unless noted, sampling effort did not vary with burn category.Species resolutionMost entries in the species list represent ontogenetic stages or constituent parts. With the exception of a few nodes (i.e. algae, moss, lichen, mycorrhizae, saprophytic bacteria, saprophytic fungi and nematodes) this ecological resolution is consistent throughout the list. Taxonomic information was assigned using the Global Biodiversity Information Facility database27, but taxonomic resolution varies. Vertebrates are identified at the species level. Invertebrates, while distinguished as morphospecies, were not identified below the family level in many cases. Despite not always being identified to the lowest taxonomic category, entries represent life stages of distinct species or groups, and are accompanied by all the taxonomic information we could provide at that level of resolution.Biomass density estimationEach species observed in our sampling methods received a biomass density estimate. Biomass density was estimated as the product of mean individual body mass and density. We estimated body-mass either by weighing individuals directly, or conservatively estimating their mass volumetrically. For the latter, we measured the length (tip of abdomen to tip of head), width (max width of body excluding appendages) and depth (max depth of body excluding appendages) of individual organisms and converted into an approximate volumetric shape (e.g. ellipsoid, cylinder, hemisphere, etc.)28,29. Mass was estimated by multiplying this volume by a tissue density of 1.1 g/mL30,31,32. Body sizes for stages that were not directly observed were inferred from published records and databases. Density estimates were derived from the sampling methods discussed below and varied with burn category. Mean and standard errors were derived from the nine replicate sites within each treatment, unless otherwise stated. Due to the brief sampling window at each site, these estimates should be regarded as point estimates rather than integrated over time.Plant samplingVegetation transectsTo estimate density and biomass of trees, shrubs, and ground cover, we conducted vegetation transects at each site. Transects were located within each 200 m × 200 m plot but were offset from other sampling methods by 10 m to avoid trampling. Transect direction was chosen at random and two transects were run in parallel. Each transect was 50 m in length but width (and distance between transects and sampling area) varied with sampling method as detailed below.Ground coverTo estimate plant ground cover, we surveyed 1 m2 quadrats at 5 m intervals on each 50 m transect. Summed, this gave a pooled ground cover sample area of 20 m2 at each site. Within each quadrat we estimated the percent cover and average height of grasses, forbs, woody litter, soft-loose and soft-rooted litter, shrubs, and trees. We identified small trees, shrubs and dominant forbs (i.e., forbs with the greatest cover at a site) to species. We then used cover area and height to estimate a volume for each species or group. Volumes were later multiplied by taxon-specific measurements of mass-to-volume ratios derived from vegetation box quadrats to obtain biomass estimates for each ground cover species or group.Canopy coverTo estimate canopy cover, above each of the 20 ground cover sampling locations, we identified each tree species overhead and estimated its absolute canopy cover. This gave a pooled canopy cover sample area of 20 m2 at each site. Canopy cover estimates were incorporated into arthropod biomass density estimates from fogging.TreesTo estimate tree density, we identified and measured all trees that exceeded 15 cm diameter-at-breast-height (DBH) within a 15 m wide band extending the length of each 50 m vegetation transect. Small trees (DBH 1 cm. We estimated shrub volume (length × width × height) and biomass density by combining volume and density estimates with taxon-specific measurements of mass-to-volume ratios derived from vegetation box quadrats. We estimated above ground biomass for small trees using species-specific DBH-to-biomass conversions33.

Understory plants
To estimate the volume and cover of understory vegetation we used three-dimensional box quadrats. These box quadrats consisted of a canvas-walled rectangular box with the floor panel removed, that was placed over the understory to be surveyed. The canvas walls prevent arthropods from escaping as they were also used in arthropod sampling. Each box quadrat measured 1 m × 0.5 m × 0.5 m, covering a ground area of 0.25 m2. At each site, we placed box quadrats at three locations representative of the dominant vegetation. This gave a pooled sample area of 0.75 m2 at each site. To generate a volume estimate, within each box we estimated the percent cover and maximum height of each plant species. Leaf litter was treated separately but also quantified. Then, using a Velcroed flap as access (designed to prevent bugs from escaping), we collected and weighed all the above ground plant biomass, separated by species and type (litter, branches, leaves, etc.). A subsample of material (up to 1 kg) from each plant species and type was retained and returned to the station to extract the associated arthropods via Berlese funnels (detailed below).
For each plant species measured in the vegetation box samples, we estimated a mass-to-volume ratio. We estimated the biomass of understory plants by combining our mass-to-volume ratios with volume estimates from ground cover transects. For taxa present in ground cover transects but not vegetation boxes, or with fewer than three measurements of mass-to-volume ratio, we pooled measurements from higher taxonomic levels. For example, if there was only one measurement of species A, but two measurements for a congener, species B, we used measurements for both species A and B to represent the mass-to-volume ratio of each. In this way, we estimated mean mass-to-volume ratios for every species-type encountered in the shrub and cover transects.
Coarse woody debrisTo estimate the mass density of woody detritus we quantified it according to the method detailed in Waddell (2002). Specifically, we used the center of the vegetation transect as a line-intercept for woody detritus. Woody debris was recorded if its longitudinal axis intersected the transect line, its diameter at the point of intersection was ≧ 12.5 cm; it was ≧ 1 m long and it was not decayed to the point of disintegration. For each piece of woody detritus, we measured the diameter at both ends, the length, and the stage of decay. Length was measured only for the portion exceeding 12.5 cm in diameter. We estimated the volume (m3) for each piece of debris:$$V{m}^{3}=left.frac{(pi /8)({D}_{s}^{2}+{D}_{L}^{2})l}{10,000}right)$$
(4)
({D}_{S}^{2}) is small diameter (cm), ({D}_{L}^{2}) is the large diameter (cm) and l is the length (m)34. We converted this estimate to a volume-density (m3 ha−1) estimate:$${m}^{3}h{a}^{-1}=left(frac{pi }{2;{L}_{t}}right)left(frac{V{m}^{3}}{{l}_{i}}right)f$$
(5)
Lt is the combined transect length (100 m), lt is the length of the individual piece, f is a conversion for area (10,000 m3 ha−1)34. Finally, we converted this is a dry-weight biomass density (kg ha−1):$$kg;h{a}^{-1}=({m}^{3}h{a}^{-1})left(frac{1000,kg}{{m}^{3}}right)SpGast D$$
(6)
SpG is a specific gravity estimate and D is the correction for the state of decay34. Specific gravity can be a species-specific estimate. Woody debris is difficult to identify to species so we applied a single specific gravity estimate to all debris weighted by the relative abundance of tree species in our surveys. We used a weighted specific gravity of 0.382, estimated by multiplying the mean specific gravity of Pinaceae (0.372) by their relative abundance (0.947) and adding that to the specific gravity of Quercus (0.566) multiplied by their relative abundance (0.053). We applied the weighted means to the decay corrections in Waddell (2002).Species of uncertain identificationSome plant identifications were difficult, particularly at burned sites, where some individuals were identified to genus or family. To assign species identities to individuals classified a higher taxonomic rank at a focal site, we first randomly selected a proxy site from the nine unburned sites. This proxy approach is possible because the composition of burned and unburned sites was similar before the fire. Proxy sites were randomly selected from among all the unburned sites because the spatial grouping didn’t lend itself to maintaining distinctions between blocks. Using the species abundance distribution of the focal plant taxon at the proxy site, we randomly assigned (with replacement) plant species identities to unidentified individual plants at the focal site. We repeated this random sampling 999 times, then estimated the resulting mean species abundance and bootstrapped standard errors at the treatment level.Invertebrate samplingGiven the diversity of invertebrates in the ecosystem, and the biases associated with each invertebrate sampling method, we employed several methods to quantitatively survey the entire community.Arthropods on understory plantsTo estimate the density of arthropods on understory plants we collected them along with plant material in the box quadrats. Up to 1 kg of plant material of each species was bagged inside box quadrats and transported to the field station for processing. Insects were extracted from the plant material in Berlese funnels that were hung in the shade at ambient temperatures and powered with 60 W frosted-incandescent bulbs. All plant samples were processed in funnels for 72 hours, after which they were checked twice a day (collecting jars changed). After no additional arthropods had been collected for 24 hours samples were removed. Arthropods were placed in ethanol for later identification. After identification, arthropod densities were estimated as the product of the arthropod-to-volume ratio for each plant and the total volume of each plant estimated from transects. We collected 4543 individual arthropods from 170 morphospecies associated with plant material from box quadrats, processed in Berlese funnels.Soil-dwelling arthropodsTo estimate the density of soil-dwelling arthropods we collected soil cores. Alongside each box, we collected four cores, each 10 cm in diameter. Two shallow cores were sunk to a depth of 5 cm, giving a combined (6 replicates total) sample area of 0.0471 m2 at each site. Shallow cores, which targeted small arthropods (e.g. collembola, diplura, acari), were collected and transported back to the field station for processing in Berlese funnels. Soil samples were processed in Berlese funnels in the same manner as plant material. Two deep cores were sunk to a depth of 15 cm, giving a combined (6 replicates total) sample area of 0.0471 m2 at each site. Deep cores targeted large invertebrates (e.g. lumbricidae, myriapoda) and were processed by hand in the field. Any large invertebrates encountered were placed in ethanol for later identification. After identification, densities were estimated as the quotient of counts and area sampled. We collected 690 individual arthropods from 57 morphospecies from soil cores.Arthropods on hard substrateTo estimate the density of arthropods like wasps and spiders on tree trunks, rock and logs we visually surveyed these hard substrates. In three locations at each site, we surveyed all hard substrates within a cylinder 5 m in diameter and 2 m in height. The total sample area at each site was 58.9 m2. All invertebrates larger than 2 cm in length were collected and fixed in ethanol for later identification. After identification, hard substrate densities were estimated as the quotient of counts and sample area. Hard substrate surveys yielded 616 individual arthropods from 74 morphospecies.Understory arthropodsTo estimate the arthropod densities associated with understory shrubs at each site we supplemented box quadrats with sweep net surveys. We used nets with a 38 cm diameter opening (Bioquip) to sweep the same 5 m diameter area as the hard substrate surveys. Three replicates at each site gave a total sample area of 58.9 m2. Net sweeps were performed at a constant pace of 2 arcs per meter. All invertebrates collected were fixed in ethanol for later identification. After identification, arthropod densities were estimated as the quotient of counts and sample area. Sweep net surveys yielded 706 individual arthropods from 168 morphospecies.Canopy arthropodsTo survey arboreal arthropods, we used a thermal fogger to loft insecticidal fog into the tree canopy. While this method can sample a large area, we were hampered by permit and accessibility issues at some sites. As a result, we sampled a representative subset of 12 sites (6 unburned, 3 high severity, and 3 low-to-moderate severity) with canopy fogging. Chemical fogging remains the most widely used method for sampling canopy arthropods35,36,37. It is notoriously difficult to assess canopy arthropods with any other method35. We used an IGEBA TF-35 Thermal Fogger to disperse a pyrethrum-based insecticide (EverGreen Crop Protection EC 60-6) into the forest canopy. During each fog we dispersed 4 L of solution (a 7% concentration of insecticide in a water-based carrier-dispersant) in 10 minutes over two representative trees and one live shrub at each site. Live trees were sampled at unburned sites, dead trees were sampled at high severity sites, and one dead and one live tree were sampled at low-to-moderate severity sites. Fogging was done under low-wind conditions allowing the thermal fogger and temperature gradients to lift the fog into the canopy. White 2.25 m2 tarps were placed beneath the fogged area to collect insects. The number of tarps used varied with the size and shape of the canopy. After an optimal elapsed drop-time of 120 minutes37, arthropods were collected from tarps and placed in ethanol for later identification. To estimate site-level arthropod abundance using canopy fogging, we estimated the density per square meter of tarp for each morphospecies and habitat type (living tree, dead tree, or shrub), then multiplied these densities by the total area covered by each habitat type. The area covered by each fogged habitat type was obtained from canopy cover transects (trees) and shrub transects (shrubs). Mean and standard errors for density estimates were estimated using the 3 (or 6) sites per treatment. Canopy fogging yielded 16,353 individual arthropods from 489 morphospecies.Strong-flying insectsTo estimate the diversity and relative abundance of strong-flying arthropods, we utilized blacklight traps (Miniature Downdraft Blacklight (UV) Trap Model 912). We supplemented our sampling with blacklight traps, because strong flying arthropods (e.g. vespid wasps) may not be adequately sampled by fogging or sweep netting. Black lights were deployed at sites prior to dusk on the third night of bat acoustic surveys (see below). Each site received one trap and black lights were not deployed on rainy or windy nights. Samples were collected the following morning. To preserve their integrity for later identification, lepidoptera samples were removed and frozen, other flying insects were fixed in ethanol. Black light traps collected 5,508 individuals from 279 morphospecies.Black light samples were only used to estimate densities for those species not regularly captured with other methods. Two hundred fifteen morphospecies were only found in black light samples. Black light traps, which can sample large areas, are an efficient means of generating diversity estimates for nocturnal flying insects. However, black light traps generate activity densities rather than absolute densities, over sample areas that vary within and between nights. To convert counts to density estimates for morphospecies collected only in black lights we paired them with an analogous species (based on taxonomy and body size). These density analogues were captured at the same site in both black lights and at least one other method with explicit absolute densities. We estimated the mean relative abundance ratios of these species in black lights across all sites in which they co-occurred. We then multiplied the density of the ecological analog by these abundance ratios to estimate densities of all species that were collected only in the black light traps.Larval biomass densityTo estimate the density of larval Formicidae, lepidoptera and coleoptera, we partitioned them according to the relative abundance of their adults present in the same treatments. Larvae for many hemimetabolous invertebrates can be difficult to identify without molecular methods. When larvae could not be identified to species, they were binned into categories of Formicidae, lepidoptera and coleoptera. We then collected body size information and density estimates for these stages. Larval densities were then partitioned according to the relative abundance of adults within each treatment. Adults were eligible to receive larvae if (1) they belonged to the same taxonomic rank as the larvae, (2) they did not already have any observed larval stages, (3) there were available resources for their larvae in the treatment.Vertebrate samplingSmall mammalsTo estimate the density of mammals less than 2.5 kg we used live traps uniformly distributed over a 90 m × 90 m grid38. We placed 100 traps 10 m apart, alternating between large (7.5 cm × 9 cm × 23 cm) and extra-large Sherman traps (10 cm × 11.5 cm × 38 cm)39. Traps were baited (a mix of oats, peanut butter, bird seed and molasses) and covered with natural materials for insulation. Traps were locked open and pre-baited for three days and then operated for three consecutive nights. Traps were closed during the day due to high daytime temperatures and lack of shade, particularly in the high severity burn areas. Traps were re-opened in the late afternoon. Captured mammals were identified to species (or individuals during recaptures), weighed, measured and fit with ear tags for future identification, then released. In 7,906 trap nights (adjusted for trap failures), we had 988 captures of 11 species of small mammals.Live-trapping and marking of small mammals allowed us to estimate their abundance using spatial capture-recapture (SCR) models40. These models use individual detections in combination with detection locations (here, location of a given live trap within the grid) to estimate density while accounting for imperfect detection and variation in detection due to variability in individual exposure to the trapping grid. When data are collected across several trapping grids, as in the present study, the joint modeling of all data allows estimation of grid-level covariate effects on density41. We used this framework to analyze data from all 27 plots jointly and allowed for density to vary according to burn category of a site (unburned, low-to-moderate severity or high severity), so that, for example, all unburned sites had the same density of a given species. If a species was never caught in a burn category, we fixed its density (and consequently, its biomass) to 0 for all sites in that burn category.Because trapping data for most species was too sparse to fit species-specific models, we grouped ecologically similar species and built models that shared parameters among grouped species. In this manner, we analyzed the joint data of mice and chipmunks: Brush mouse (Peromyscus boylii), Deer mouse (Peromyscus maniculatus), Pinyon mouse (Peromyscus truei) and Western harvest mouse (Reithrodontomys megalotis), Yellow-pine chipmunks (Tamias amoenus), Long-eared chipmunks (Tamias quadrimaculatus) and Shadow chipmunks (Tamias senex).In the mouse model, Brush mouse and Pinyon mouse densities were assumed to have the same response to burn category (but not the same densities), whereas Deer mice were allowed a different response to burn category. This choice was made based on capture frequencies of species in the different burn categories. Pinyon mice and Western harvest mice had a fixed density of 0 in low-to-moderate and high severity burn sites. All species shared the same detection parameters.Even combined, the chipmunk data captures were too sparse to estimate species specific densities. We therefore treated all chipmunks like a single species to estimate overall chipmunk density, then calculated species specific densities by multiplying overall density with the proportion of individuals of a given species in the data. For example, if species A made up 50% of all individuals in the data, we would calculate density for species A by multiplying overall chipmunk density by 0.5. This model entails the assumption that the effect of burn category on density was the same for all chipmunks, which seems reasonable based on capture frequencies.We built species-specific SCR models for California ground squirrels (Otospermophilus beecheyi) and Dusky-footed woodrats (Neotoma fuscipes); for the latter we fixed density to 0 in high severity sites. Northern flying squirrels were only caught 3 times at unburned sites, so we set its density to 0 in low-to-moderate and high severity sites, and used a published density estimate from a Sierra Nevada site42 for unburned sites. Because individuals of Trowbridge’s shrew (Sorex trowbridgii) were never recaptured and had high rates of trap mortality, we estimated their abundance using non-spatial removal models43 for unburned and moderately burned sites and set their abundance to 0 at high severity sites. We transformed abundance to density by dividing it by the 8100 m2 area (90 × 90 m) of the live trapping grid.We fit SCR models in R using the package “secr” ver. 3.1.344, and removal models using the package RMark ver. 2.2.445.BirdsTo estimate the diversity and density of birds, we conducted point count surveys46. Each site had two point-count stations, spaced at least 200 m apart, that were surveyed on the same day. Counts were conducted at each site on three different days per season. To mitigate migration effects all counts were conducted between mid-June and mid-July. Bird surveys did not take place when it was raining, extremely cold (20 kmh). Wind speeds were obtained with handheld anemometers. Point counts began 15 minutes after sunrise and were completed by 10:00 AM, corresponding to when passerine birds are most active. Each survey consisted of one ten-minute count, split into two five-minute periods. Each individual bird was recorded only once over the entire count. Trained observers identified all birds to species by sight or sound and estimated the number of individuals of each species within 100 m of the sampling point. This gave us a sample area of 15,708 m2 at each site. Birds that flew through or were detected outside of the survey area or survey time were documented but not included in richness or density estimates. During 162 point-surveys we observed 1,039 birds from 52 species (107 individuals could not be identified).The repeated point counts allowed us to estimate bird abundance and density using N-mixture models47. These models use repeated counts of individuals to estimate abundance within a sampling unit (in this case, the 100 m radius point count) while accounting for imperfect detection. Because abundance estimates refer to a set area, these can be converted to densities. Because data for some bird species were sparse, we fit data of all species jointly in a community model48. In community models, each species has its own set of parameters, but species-specific parameters are modeled as coming from a common underlying distribution that is shared by the community of species (essentially, a species-level random effect). This constitutes a form of information sharing, which improves parameter estimates for data-sparse species. In our community N-mixture model, we allowed for species-specific detection probabilities, as well as species-specific effects of burn category on abundance. We further included a fixed (across species) effect of the amount of wind during a given survey on detection, as wind can impair auditory detection of birds. We implemented the community N-mixture model in a Bayesian framework using the program JAGS ver. 4.2.049 accessed through R with the package jagsUI ver. 1.5.050. JAGS fits models using Markov chain Monte Carlo (MCMC) algorithms; we ran 100,000 MCMC iterations and discarded the first 50,000 as burn-in. We checked for chain convergence using the Gelman-Rubin statistic51; all parameters had a value 2.5 kg we used camera traps. At each site, we deployed two Reconyx HC600 Hyperfire cameras, which have a no-glow infrared flash, preventing disturbance to wildlife and detection by humans. To reduce false triggers, we set cameras in shaded locations and cleared vegetation in front. Cameras were deployed along landscape features that suggested animal movement: game trails, forest openings, abandoned dirt roads, etc. Cameras were set at least 50 m apart and strapped to trees 40 cm above the ground and 1–2 m away from the edge of a trail or opening. Cameras were operated for 24 hours a day, with 3 consecutive pictures per trigger and no time lapse between triggers. All 54 camera traps were installed by early-July and retrieved in mid-September. Cameras were checked after four to six weeks for SD card and battery replacement. All pictures were reviewed manually for species identification; identified pictures of animals were organized into camera and species-specific folders for post-processing in the R computing environment ver. 3.4.352 using the package camtrapR ver. 0.99.553. Even though some bird and small mammal species were occasionally photographed by camera traps, we excluded their photo-records from further analysis. After adjusting for malfunction, cameras operated for 4,238 trapping nights over which they assembled 12,243 independent records (detections that were at least one hour apart if of the same species at the same camera) comprising 10 species of mammals >2.5 kg.Because camera-trap images do not allow identification or counting of individuals for analysis with SCR or N-mixture models, we used the Royle-Nichols (RN) occupancy modeling framework54 to estimate abundance of medium/large mammal species. Occupancy models55 use repeated species detection/non-detection data from a collection of sampling locations to estimate species occurrence probability while accounting for imperfect detection. The RN model makes use of the fact that the probability of detecting a species at a site increases with the abundance of that species at that site, allowing estimation of site-level abundance from species level detection/non-detection data. We used the R package camtrapR53 to convert raw camera trap data for each species into a binary (detected = 1, not detected = 0) location-by-occasion format. Because of their proximity, we considered both cameras at a given plot to constitute a single sampling location. We defined an occasion as 10 consecutive days of sampling. To account for malfunctioning of some cameras, we calculated effort as the number of days per occasion that each pair of cameras was functional.Because data for some species were sparse, we jointly analyzed data for all species in a community RN model (see Birds for a description of community models); we allowed for species-specific detection probabilities as well as species-specific effects of burn category on species abundance and included a fixed effect (across species) of effort on detection. The RN model returns sampling location level estimates of abundance, but in contrast to bird and reptile surveys, which were explicitly linked to a specific sampled area, the area sampled by a camera trap is not easily defined. Mammals recorded in the relatively small detection zone of a camera use much larger areas. We calculated densities of large mammals by dividing the abundance estimates from the RN model by the average home ranges of the recorded species. We fit the community RN model in a Bayesian framework as described for birds, running 30,000 MCMC iterations and discarding the first 15,000 as burn-in. All chains converged, according to the Gelman-Rubin statistic.To obtain estimates for home range size and individual mass of large mammals, we searched the scientific literature for studies conducted in similar habitat (temperate coniferous forests) and for home ranges, during the summer and fall seasons. When no information from a similar habitat was available, we used studies from forested areas. When no information for target species was available, we used information from congeners. When multiple studies were available, we calculated the mean across studies, weighted by sample size if provided. Similarly, when studies provided information separately for males and females, we calculated a weighted mean. For some large mammal species, we only found information on annual home ranges, and to approximate a summer/fall home range, we multiplied these annual ranges by 0.67.BatsTo survey bat community composition and abundance, we conducted acoustic sampling at each site. We placed a Wildlife Acoustics Songmeter SM4BAT FS echolocation detector at each site for a minimum of three consecutive nights. The detectors were set to record from sunset to sunrise when bats are most active. We attached microphones to t-posts placed in open areas, elevating them 2.5 meters from the ground to minimize signal attenuation from nearby vegetation and canopy. Bat calls were analyzed using SonoBat software to assess likely species for each file. These identifications were vetted by an experienced bat researcher, Ted Weller (USDA Forest Service). One site (B3U1) was not sampled with bat detectors, whereas two sites had one and two additional nights sampled, respectively. During 81 recorder nights, acoustic recorders captured 17,484 calls, of which 7348 were identified to 17 species.Similar to camera traps, bat calls recorded by acoustic recorders do not allow counting or identifying individuals. Therefore, we built a community RN model to estimate abundance for bats. The model structure was identical to that described for medium/large mammals, except that it did not include any covariate on detection probability. Analogous to our medium/large mammal analysis, we used literature information on average home range size to convert bat abundance to bat density. We ran the bat model for 150,000 iterations and discarded the first 75,000 iterations as burn-in. All chains converged according to the Gelman-Rubin statistic.ReptilesTo estimate community composition and density of reptiles, we conducted timed searches within the 90 m × 90 m small mammal trapping grids at each site. This gave us a sample area of 8100 m2 at each site. Reptile surveys were conducted either prior to small mammal trapping or two weeks post to minimize impacts from sampling disturbance. Reptile surveys were conducted between 8:00 am and 10:00 am by teams of two to four, for a total of one person-hour. We performed reptile surveys three times at each site, with at least one week between surveys. During searches, all refugia (i.e., rocks, logs) within the plot were carefully overturned and then replaced. In addition to reptile species, we documented time of day and weather, as well as cover type and cover length if applicable. Our reptile surveys yielded 145 records of two snake species and four lizard species. Two additional snake species were observed incidentally during other sampling methods.Repeated counts of reptiles allowed us to use N-mixture models47 to estimate abundance and density. However, reptile data were extremely sparse and contained few species, precluding the use of a full community model as fit for birds. Instead, we structured reptile data into three groups: snakes (Western terrestrial garter snake (Thamnophis elegans) and Yellow-bellied racer (Coluber constrictor)), Alligator lizards (Elgaria coerulea and E. multicarinata) and Western fence lizards (Sceloporus occidentalis). We then combined data from all groups in a single N-mixture model, allowing for group-specific abundances. Due to sparse data, we only estimated an effect of burn category on abundance for fence lizards. Abundance for other species groups was assumed to be constant across burn categories in the model. To obtain species and burn category specific estimates of abundance, we combined model estimates of abundance with raw counts of individuals. For the two snakes, we calculated species level abundance by burn category by multiplying overall model-estimated snake abundance with the proportion of individuals made up by each species in each burn category. Because some Elgaria sp. observations could not be identified to species level, we did not attempt to calculate species specific abundances (but we note that the majority of observations was of E. coerula). We calculated species/group density by dividing abundance by the size of the sampling unit, in this case, a 90 × 90 m square. We fit the N-mixture model in R using the package unmarked ver. 0.12.256.Species identificationVertebrates were identified in the field or from photographs. Invertebrate specimens were collected and fixed in the field for identification in the lab where they were split into morphospecies using published keys57,58 and consultation with experts at the UC Davis Bohart Museum. Morphospecies were not always identified to lower taxonomic levels but were distinguished from similar species by coarse morphological characters. Prior to density estimation some morphospecies were aggregated based on taxonomy, body size, sample methodology, co-occurrence patterns and the difficulty in distinguishing members of the group. The rationales behind aggregations are documented in the raw invertebrate sampling data.Body size estimationFood webs are commonly organized around body size59,60,61,62,63 and we include length and mass estimates for all life stages in our data set. We assume mean body size does not change substantially across burn severities and use a single estimate for all life stages regardless of treatment. When possible, we measured 10 individuals for each morphospecies and life stage (for invertebrates and small mammals only, as individuals from other taxonomic groups were not handled). Body mass for small mammals was measured directly with a Pesola scale. Body length estimates for invertebrates were derived from the longest measurement. Volumetric estimates were derived by applying the measured length, height and width to the three-dimensional shape that most closely approximates that of the organism (e.g. ellipsoid, rectangular prism, cylinder, hemisphere, cone). We estimated biomass by multiplying these volumetric estimates by a tissue density of 1.1 g/mL30.Body sizes for organismal life stages observed but not captured (e.g. adult birds and large mammals) were estimated from the literature. Life stages for many organisms were not directly observed (e.g. lepidoptera eggs) but are almost certainly present. For these stages we again used the literature to estimate the body sizes. For example, we used a data set of egg sizes from 6,700 insect species to inform estimates for the species in our study64. The species list indicates the estimation method and literature source used to estimate the body size of each individual life stage is included in the species list.LinksEcological networks can be broken into two types: Undirected and directed. Undirected networks include bipartite networks (e.g. plant-pollinator, host-parasite) and social-networks. In undirected networks, interactions (links) and their participants (nodes) are observed at the same time. Links are not inferred in undirected ecological networks (unless false-negatives due to sampling error are taken into account) because they are directly observed (e.g. tick removed from a lizard). Undirected links can be weighted (interaction strength) by counting observations.Directed networks include food webs. Most food web links are inferred because it is not feasible to evaluate all possible consumer-resource interactions in a system through direct observation. The number of possible consumer-resource interactions in a food web is equivalent to the number of nodes squared. There were 3,084 nodes in the Eldorado National Forest, resulting in 9,511,056 potential consumer-resource interactions. In the scope of an ecological study, it is rarely feasible to directly observe most feeding links for most species15, much less an entire web, though molecular methods are bringing this closer to possibility for some guilds (e.g. large mammalian herbivores)65. While often detailed for birds and mammals, published accounts of direct observations of diets are often general (e.g. “Species A eats organic matter”) or entirely lacking for other groups. Restricting link assignments only to those that are directly observed will not create an accurate or unbiased food web.To assign resources to consumers we supplemented our own field observations with published diet records, expert opinion and rule-based filters. In the absence of direct observation, rule-based filters are an important tool for sorting through the huge number of potential feeding interactions. Filters varied with the species and ontogenetic stages to which they were applied but consisted of two main types: Encounter and compatibility. Encounter filters determine the potential for consumers to interact with resources. Encounter filters are based on habitat co-occurrence, forging strategy and diel activity patterns. Applied to links that have passed through encounter filters, compatibility filters determine the outcome of a potential encounter. Compatibility filters are based on consumer diet information, consumer diet breadth, resource palatability, and consumer-resource body size ratios. For inclusion in the food web, a link must pass through both filters.Link assignmentAll species were considered potential resources. For feasibility and reproducibility, species were broken into groups of encounter filters. First, we separated plants, fungi and detritus from other metazoans.Fig. 2Sampling sites map and King Fire perimeter. We sampled 27 sites total in the Eldorado National Forest, California, three years after the King Fire, nine in each burn category: Unburned, moderate severity, and high severity. Features not indicated in legend are typical of topological maps.Full size imageFig. 3Representative map of sampling design. We employed 19 different methods to estimate the richness and biomass density of organisms in the Eldorado National Forest, California, three years after the King Fire. Methods were conducted entirely within or centered within the 200 × 200 m site perimeter. Methods were paired in space and time when useful (e.g. black lights and acoustic bat surveys), and separated when necessary to avoid interference (e.g. small mammal trapping grid and vegetation transects). All methods at a site were conducted over 4–5 consecutive days.Full size imageThe first encounter filter was a loose group consisting of primary producers, non-living resources (e.g. detritus, carcasses), and saprophytes. In this first filter, primary producers were broken into parts (e.g. root, seed, leaves, etc). With few exceptions (i.e. moss, lichens, algae) primary producers in this filter group were evaluated as families or species. Saprophytes were evaluated as spores or adults. These groups served as the encounter filter for primary consumers and fungivores.Next, we partitioned all other metazoans into smaller encounter filters based on phylogeny, behavior, activity, habitat and palatability. A species’ life stage can be a member of multiple resource groups. These groups served as the encounter filter for most predatory or omnivorous organism stages. These encounter filters included:

Flying invertebrates

Nocturnal flying invertebrates

Diurnal flying invertebrates

Non-flying invertebrates

Ground-dwelling invertebrates

Ground-dwelling invertebrates excluding spider eggs

Soft-bodied ground dwelling invertebrates

Invertebrates on plants

Invertebrates on plants excluding spider eggs

Soft-bodied invertebrates on plants

Invertebrates on trees

Reptiles

Birds

Small mammals

Large mammals

Encounter filter exposure was tailored to consumer type. For consistency, consumers were broken into phylogenetic and ontogenetic guilds (e.g. lepidoptera larvae). For example, web-building spiders were exposed to flying insects, but not small-mammals encounter filters. Compatibility filters are specific to species and stage. For example, while both are exposed to the flying-insects encounter filter, adult and juvenile web-building spiders will have different compatibility filters because they have different consumer-resource body size ratios. Phylogenetic and ontogenetic consumer guilds are detailed below, but the decisions for each of the 178,655 links assignments in the food web can be found in and reproduced with the R code accompanying this manuscriptSpidersAs common, generalist insectivores, spiders are important consumers in the network66. To assign them resource links, spiders were broken into three ontogenetic stages: adult, egg, juvenile. Spiders were then separated into 17 consumer guilds by Family. The 87 spider species that could not be identified at the family level were assumed to be web-building spiders. To assign feeding interactions, we applied one or more invertebrate resource group filters (see above) to the members (species and stages) of each spider guild. Next, each link passed through a consumer-resource body size filter before being included in the network. For example, web-building spiders were assigned to capture flying insects and consume insects larger than 20% of their own body-length (as mesh size sets a lower limit of prey size) and less than 200% of their own body length67. Because of the generality of their filter-feeding hunting strategy, web-building spiders accounted for nearly half (32,663) of interactions with spiders as consumers. Additionally, when it was reported in the literature, adult members of a guild were assigned to consume nectar.Herbivorous beetlesTo assign consumer links to non-predatory beetles, their larval and adult stages were separated into Family-level feeding guilds. When information was available, links were evaluated at the species-stage level rather than the Family level. Resource links containing primary producers and non-living resources (detritus, carcasses, dung, etc) were assessed based on direct observation, published reports and expert opinions. When feeding on living plants, beetles were assigned to specific tissue types (e.g. flower, leaves, roots). For example, Cerambycidae_sp_2 was assigned to feed on the stems of Arctostaphylos patula as well as stems of plants from the family Rhamnaceae. This assigned 10,289 consumer links to herbivorous beetles.Larval lepidopteraCaterpillars were the most speciose herbivore group in the network. Larval lepidoptera vary greatly in host plant range and the quantity of information on their feeding habitats and were not lumped into guilds whenever possible. Microlepidoptera were not identified below the order level and were assumed to feed like Gelechiidae caterpillars (many species of which have been documented in Eldorado National Forest). We assigned links to caterpillars with resource filters of host plants and parts based on published diet records, palatability, expert opinion and observed co-occurrence between potential hosts and adults. This assigned 2,434 consumer links to caterpillars.Adult lepidopteraAdult lepidoptera feed primarily or exclusively on nectar from flowering plants. The host ranges for most adult lepidoptera are not well described. To assign nectar links, adult lepidoptera were grouped into families and passed through resource filters based on published records, co-occurrence, and expert opinion. Non-feeding adult moths were removed from link consideration. These filters assigned 5,854 nectar links to adult lepidoptera.Adult hymenoptera as herbivoresMost species of hymenoptera are parasitoids as larvae and free-living as adults (eusocial hymenoptera are a notable exception). Many adult parasitic wasps supplement their protein intake with nectar and pollen. The plant host ranges for solitary adult wasps are not well known. We applied nectar and pollen resource filters to adult hymenoptera based on published records, co-occurrence and expert opinion. These filters assigned 4,880 links to adult hymenoptera.Predatory beetlesPredatory and scavenging beetles are important consumers in terrestrial ecosystems. At the family level, the diets of predatory beetles are well-described relative to other arthropods. We grouped predatory and scavenging beetles into family guilds. We applied resource filters to predatory and scavenging beetles based on published records and expert opinion, habitat overlap, foraging strategy, palatability, and body size. These filters assigned 10,320 resource links to predatory beetles.FliesFlies in the Eldorado National Forest are a speciose group with diverse consumer strategies: herbivores, predators, scavengers, detritivores, parasitoids, and micropredators. Because of their trophic diversity each dipteran family was treated as its own guild, even then there were often large ontogenetic diet shifts across stages within a guild. We applied resource filters to dipterans based on published records and expert opinion of their diet. These filters were based on habitat overlap, palatability, and for predatory stages consumer-resource body size ratios. These filters assigned 9,541 resource links to flies.Hymenopteran parasitoidsParasitoid wasps are a diverse group that can have strong, even regulatory, effects on their host populations. We applied resource filters to hymenopteran parasitoid larvae in a three-step process. First, we used published records and expert opinion to establish their host range and host specificity. Next, we looked for the subset of potential hosts that co-occured with adult parasitoids across the most sites. Finally, we retained the subset of those host species whose adult body sizes were equivalent to or slightly larger than adult parasitoids. These filters assigned 4,695 host links to parasitoid wasps.Other hymenopteraThe remaining hymenoptera were composed of wood wasps and gall wasps, as well as bees, predatory wasps and ants. Host plant links were assigned to wood wasps and gall wasps in the same way as larval lepidoptera. Nectar links were assigned to bees in the same way as adult lepidoptera. Prey links were assigned to solitary and eusocial wasps in the same way as insectivorous gleaning birds. Ants were treated as generalist omnivores whose links were assigned links in a manner similar to all of the above.HemipteraHemiptera were the most abundant consumer group in the Eldorado National Forest. Hemiptera use their proboscis to feed on fluids, either as herbivores or predators. To accommodate this variation in consumer strategies resource filters were assigned to individual hemiptera species based on phylogeny. Host plants were assigned to herbivores based on field observations, published records and expert opinion. Plant fluids (with the exception of sap from soft woods) were not included in the node list, so herbivores were assigned feeding links based on the plant tissues that they pierced (i.e. stem or leaves). Predatory hemiptera were assigned insects on plants passed through consumer-resource body size filters. When more specific dietary information was available resource filters were further refined. These filters assigned 3,177 links to the hemipteraCollembolaFrom soil to canopy, springtails were widely distributed across Eldorado National Forest habitats and are important resources for arthropod secondary consumers. Collembola are herbivores, detritivores and fungivores. Little diet information is available for collembola, which were assigned resource filters based on habitat and phylogeny (order). Habitat was determined by collection method, and resource availability was determined by habitat. These, phylogenetic-habitat-resource filters assigned 140 links to the collembola.Bark licePsocoptera were common on plants in the Eldorado National Forest, feeding on algae, lichen, moss, fungi and detritus. They exhibit little variation in diet and were assigned resource filters as a group at the Order level. This small set of resource filters assigned 120 links to the psocoptera.ThripsThysanoptera were common and abundant consumers in the Eldorado National Forest, whose diets can range from herbivory to facultative predation and predation. To accommodate this variation in consumer strategy, resource filters were assigned to individual thysanoptera species based on phylogeny. Host plants for herbivores were based on field observations, supplemented with published records and expert opinion. Thysanoptera predators were assigned soft-bodied insects on plants passed through consumer-resource body size ratio filters. These resource filters assigned 924 links to thysanoptera.BatsBats are important consumers of arthropods. Arthropod encounter filters were determined by bat foraging strategies reported in the literature. Juvenile bats were treated separately. Aerial hawking bats capture nocturnal flying insects. Gleaning bats capture insects on plants. Pallid bats (Antrozous pallidus) which fly close to the ground, capture arthropods on the ground and understory plants. These arthropod filters then passed through consumer-resource body size ratio filters, which assigned 4,621 links to adult bats.BirdsBirds were the most speciose vertebrate group in the Eldorado National Forest. This diversity is reflected in their diets. Bird resource filters were based on published records and expert opinion. Herbivorous birds that were reported to feed on a type of plant tissue were assumed to feed on all tissues of that type, unless the literature indicated resource specialization. For example, if a bird was reported to feed on a fruit, it was assumed to feed on all fruit in the system. Resource links for predatory and omnivorous birds were passed through consumer-resource body size ratio filters. These filters assigned 56,584 links to birds as consumers.Small mammalsMammals 2.5 kg were comparatively uncommon but important consumers in the Eldorado National Forest. Large mammal resource filters were based on published records and expert opinion. Herbivorous large mammals that were reported to feed on a type of plant tissue were assumed to feed on all tissues of that type, unless the literature indicated resource specialization. Resource links for omnivorous and predatory large mammals were passed through consumer-resource body size ratio filters. These filters assigned 840 links to large mammals as consumers.Juvenile mammalsJuvenile mammals were considered unweaned. Nursing mammals were included in the species list because they are more vulnerable than adults and are potential resources to different consumers as a result. Unweaned mammals “feed” on their mothers, who are considered their only resource link. Nursing filters assigned 38 links to juvenile mammals.ReptilesWhile not diverse or common, reptiles on the mountain slopes of Eldorado National Forest are potentially important consumers for many groups. Reptile resource filters were assigned based on published records and expert opinion. All resource links were passed through consumer-resource body size ratio filters. These filters assigned 1,487 links to reptiles.MitesMites are ubiquitous and important consumers in many habitats but are often overlooked because of their small size and taxonomic difficulty. Mites were treated as consumer groups based on taxonomy. Resources for predatory mites were based on habitat and passed through consumer-resource body size ratios filters. Resource filters assigned 2,971 links to mites.Miscellaneous arthropodsA few arthropods not discussed above were not speciose enough to be dealt with separately here. These remaining miscellaneous arthropods were treated as groups at various taxonomic levels. Encounter and compatability filters for these groups were based on published records. Encounter filters for predatory (centipedes, odonates, pseudoscorpions, solfugids, mantids, opiliones, neuroptera, embioptera, pauropoda, eusocial wasps) and omnivorous (dermaptera) groups were based on habitat and passed through consumer-resource body size filters. Encounter filters were assigned by habitat for detritivores, herbivores and fungivores (archeognatha and isopods). These filters assigned 3,157 links to miscellaneous arthropods as consumers. More

Coupled model experiments for detecting vegetation-climate feedbackWe quantified changes of near-surface (2-m) air temperature (Ta) in response to the observed NH greening for all active growing seasons during 1982–2014 using IPSL-CM. We defined the three growing seasons (spring, summer, and autumn) across the entire NH domain as periods of March-April-May (MAM), June-July-August (JJA), and September-October-November (SON), respectively. For each season, a pair of transient numerical experiments was performed by modifying LAI: a dynamic vegetation experiment (SCE) forced by annually and seasonally varying LAI from satellite observations36, and three seasonal control experiments (({{{{{{rm{LAI}}}}}}}_{{{{{{rm{CTL}}}}}}}^{{{{{{rm{MAM}}}}}}}), ({{{{{{rm{LAI}}}}}}}_{{{{{{rm{CTL}}}}}}}^{{{{{{rm{JJA}}}}}}}), and ({{{{{{rm{LAI}}}}}}}_{{{{{{rm{CTL}}}}}}}^{{{{{{rm{SON}}}}}}}) for MAM, JJA, and SON, respectively) forced by annually varying LAI for all seasons, except in the season of interest when the LAI was fixed to the climatological conditions observed during 1982–2014 (Fig. S1). For all experiments, other boundary conditions, including sea surface temperature (SST), sea ice fraction (SIC), and atmospheric CO2 concentrations, were kept consistent (Methods). Therefore, differences between SCE and the control experiments characterized the effects of the observed LAI changes on Ta (hereafter denoted as ΔTa), both intra- and inter-seasonally. Multimember paired ensembles were generated for each coupled model experiment by performing 30 repeated runs but with different initial conditions (see Methods).The capacity of the IPSL-CM GCM for simulating the seasonal variations and spatial patterns of Ta was assessed by comparing the SCE simulation results with the observation-based Ta data (Methods). Throughout most of the growing season (May to October), the SCE simulation well reproduced the increasing trend and interannual variability of the NH land mean Ta observed during 1982–2014 (Fig. S2). Observational data showed that the strongest NH warming occurred in early spring (March and April) and late autumn (November). However, the SCE simulation failed to capture the exceptionally strong warming during the transitional seasons, leading to the underestimation of the annual mean warming trend (SCE: 0.237 ± 0.024 °C decade−1; observed: 0.362 ± 0.048 °C decade−1). This underestimation stemmed from a negative bias in the increase of downwelling shortwave radiation, possibly due to an absence of short-lived forcing and bias in the cloud systems37. Overall, the SCE reproduced the geographical patterns of seasonal warming reasonably well (Fig. S3), which strengthened our confidence in the model projections. Notably, it successfully captured the observed amplified warming over pan-arctic and semi-arid regions, as well as the few cases of regional cooling, such as that over northwestern North America during MAM (Fig. S3).Intra-seasonal temperature responses to NH LAI changesFor the period from 1982 to 2014, satellite-retrieved LAI showed statistically significant increasing trends (p  0.1), strong and significant JJA cooling (−0.044 ± 0.008 °C decade−1, p  0.1) (intra-seasonal feedbacks shown in Fig. 1b). The LAI-induced JJA Ta trend was equivalent to cooling of −0.15 ± 0.03 °C in JJA over the study period, offsetting the overall SCE-simulated near-surface air warming over this period by ~12.5%. This strong JJA cooling was further supported by a significant negative correlation (r = −0.64, p  0.1) or SON (r = 0.07, p  > 0.1) (Fig. S4a, c), during which the LAI-induced changes accounted for only 1.3% (MAM) and −3.2% (SON) of the concurrent greenhouse warming. We also verified the robustness of our results by performing equilibrium experiments with an independent model, the NCAR Community Atmosphere Model coupled with Community Land Model (CAM-CLM, Methods). Indeed, this model generated a similarly strong LAI-induced cooling in JJA (−0.18 °C, p  0.1) and SON (−0.05 °C, p  > 0.1) (Fig. S5).Fig. 1: Intra- and inter-seasonal temperature responses to leaf area index (LAI) changes.a Monthly trends (shadings) of Northern Hemisphere (NH) mean LAI during 1982–2014 used as input to the seasonal simulations. The dashed curve and transparent bars indicate trends of monthly LAI and seasonally aggregated LAI values, respectively. b Linear trends of Ta driven by LAI changes within the same season (intra-seasonal) and other growing seasons (inter-seasonal). Error bars in a, b indicate uncertainty ranges [1 – standard deviation (SD)]. c Monthly trends of LAI-induced air temperature changes (ΔTa), with red and blue shadings representing positive and negative trends, respectively. The bottom panel shows the overall ΔTa trends induced by LAI changes in all growing seasons, calculated as the sum of ΔTa trends from the three seasonal runs shown separately in the above panels. ***p  More

Scott Baker, C. & Clapham, P. J. Modelling the past and future of whales and whaling. Trends Ecol. Evol. 19, 365–371. https://doi.org/10.1016/j.tree.2004.05.005 (2004).CAS 
Article 
PubMed 

Google Scholar 
Clapham, P. J. & Baker, C. S. in Encyclopedia of Marine Mammals (eds W. F. Perrin, B. Würsig, & J. G. M. Thewissen) Ch. Modern Whaling, 1328–1332 (Academic Press, 2002).International Whaling Commission. Twenty-Eighth Report of the International Whaling Commission. (International Whaling Commission, Cambridge, 1978).International Whaling Commission. Thirty-Sixth Report of the International Whaling Commission. (International Whaling Commission, Cambridge, 1986).Leaper, R. & Miller, C. Management of Antarctic baleen whales amid past exploitation, current threats and complex marine ecosystems. Antarct. Sci. 23, 503–529. https://doi.org/10.1017/s0954102011000708 (2011).ADS 
Article 

Google Scholar 
Tulloch, V. J. D., Plagányi, É. E., Matear, R., Brown, C. J. & Richardson, A. J. Ecosystem modelling to quantify the impact of historical whaling on Southern Hemisphere baleen whales. Fish Fish. 19, 117–137. https://doi.org/10.1111/faf.12241 (2018).Article 

Google Scholar 
Kemp, S. & Bennett, A. G. On the distribution and movements of whales on the South Georgia and South Shetland whaling grounds. Discov. Rep. 6, 165–190 (1932).
Google Scholar 
Branch, T. A. & Butterworth, D. S. Estimates of abundance south of 60°S for cetacean species sighted frequently on the 1978/79 to 1997/98 IWC/IDCR-SOWER sighting surveys. J. Cetac. Res. Manag. 3, 251–270 (2001).
Google Scholar 
Reilly, S. et al. Biomass and energy transfer to baleen whales in the South Atlantic sector of the Southern Ocean. Deep Sea Res. Part II 51, 1397–1409. https://doi.org/10.1016/s0967-0645(04)00087-6 (2004).ADS 
Article 

Google Scholar 
Cooke, J. G. Balaenoptera physalus. The IUCN Red List of Threatened Species 2018, e.T2478A50349982. https://doi.org/10.2305/IUCN.UK.2018-2.RLTS.T2478A50349982.en (2018).Santora, J. A., Schroeder, I. D. & Loeb, V. J. Spatial assessment of fin whale hotspots and their association with krill within an important Antarctic feeding and fishing ground. Mar. Biol. 161, 2293–2305. https://doi.org/10.1007/s00227-014-2506-7 (2014).Article 

Google Scholar 
Santora, J. A., Reiss, C. S., Loeb, V. J. & Veit, R. R. Spatial association between hotspots of baleen whales and demographic patterns of Antarctic krill Euphausia superba suggests size-dependent predation. Mar. Ecol. Prog. Ser. 405, 255–269. https://doi.org/10.3354/meps08513 (2010).ADS 
Article 

Google Scholar 
Burkhardt, E. et al. Seasonal and diel cycles of fin whale acoustic occurrence near Elephant Island, Antarctica. R Soc. Open Sci. 8, 201142. https://doi.org/10.1098/rsos.201142 (2021).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Joiris, C. R. & Dochy, O. A major autumn feeding ground for fin whales, southern fulmars and grey-headed albatrosses around the South Shetland Islands, Antarctica. Polar Biol. 36, 1649–1658. https://doi.org/10.1007/s00300-013-1383-8 (2013).Article 

Google Scholar 
Herr, H. et al. Horizontal niche partitioning of humpback and fin whales around the West Antarctic Peninsula: Evidence from a concurrent whale and krill survey. Polar Biol. 39, 799–818. https://doi.org/10.1007/s00300-016-1927-9 (2016).Article 

Google Scholar 
Viquerat, S. & Herr, H. Mid-summer abundance estimates of fin whales Balaenoptera physalus around the South Orkney Islands and Elephant Island. Endanger. Spec. Res. 32, 515–524. https://doi.org/10.3354/esr00832 (2017).Article 

Google Scholar 
Meyer, B. & Wessels, W. The Expedition PS112 of the Research Vessel POLARSTERN to the Antarctic Peninsula Region in 2018. 125 p. (Alfred Wegener Institute for Polar and Marine Research, Bremerhaven, 2018). https://doi.org/10.2312/BzPM_0722_2018Knust, R. Polar research and supply vessel POLARSTERN operated by the Alfred-Wegener-Institute. J. Large-Scale Res. Facil. JLSRF 3, A119. https://doi.org/10.17815/jlsrf-3-163 (2017).Article 

Google Scholar 
Buckland, S. T. et al. Introduction to Distance Sampling: Estimating Abundance of Biological Populations (Oxford University Press, 2001).MATH 

Google Scholar 
Herr, H. et al. Aerial surveys for Antarctic minke whales (Balaenoptera bonaerensis) reveal sea ice dependent distribution patterns. Ecol Evol 9, 5664–5682. https://doi.org/10.1002/ece3.5149 (2019).Article 
PubMed 
PubMed Central 

Google Scholar 
Dorschel, B. et al. Environmental information for a marine ecosystem research approach for the northern Antarctic Peninsula (RV Polarstern expedition PS81, ANT-XXIX/3). Polar Biol. 39, 765–787. https://doi.org/10.1007/s00300-015-1861-2 (2015).Article 

Google Scholar 
Miller, D. L., Burt, M. L., Rexstad, E. A., Thomas, L. & Gimenez, O. Spatial models for distance sampling data: recent developments and future directions. Methods Ecol. Evol. 4, 1001–1010. https://doi.org/10.1111/2041-210x.12105 (2013).Article 

Google Scholar 
Hedley, S. L. & Buckland, S. T. Spatial models for line transect sampling. J. Agric. Biol. Environ. Stat. 9, 181–199. https://doi.org/10.1198/1085711043578 (2004).Article 

Google Scholar 
Hammond, P. S. et al. Estimating the abundance of marine mammal populations. Front. Mar. Sci. https://doi.org/10.3389/fmars.2021.735770 (2021).Article 

Google Scholar 
Buckland, S. T. et al. Advanced Distance Sampling (Oxford University Press, 2007).
Google Scholar 
Miller, D. L., Rexstad, E., Thomas, L., Marshall, L. & Laake, J. L. Distance sampling in R. J. Stat. Softw. 89, 1–28. https://doi.org/10.18637/jss.v089.i01 (2019).Article 

Google Scholar 
Marques, F. F. C. & Buckland, S. T. in Advanced Distance Sampling (eds S. T. Buckland et al.) 31–47 (Oxford University Press, 2004).Akaike, H. A new look at the statistical model identification. IEEE Trans. Autom. Control 19, 716–723 (1974).ADS 
MathSciNet 
Article 

Google Scholar 
Wood, S. N. Fast stable restricted maximum likelihood and marginal likelihood estimation of semparametric generalized linear models. J. R. Stat. Soc. (B) 73, 3–36. https://doi.org/10.1111/j.1467-9868.2010.00749.x (2011).Article 
MATH 

Google Scholar 
Wood, S. N., Pya, N. & Saefken, B. Smoothing parameter and model selection for general smooth models (with discussion). J. Am. Stat. Assoc. 111, 1548–1575. https://doi.org/10.1080/01621459.2016.1180986 (2016).CAS 
Article 

Google Scholar 
Dorschel, B. et al. The International Bathymetric Chart of the Southern Ocean Version 2 (IBCSO v2). Scientific Data. https://doi.org/10.1038/s41597-022-01366-7 (2022).Article 

Google Scholar 
Wilson, M. F. J., O’Connell, B., Brown, C., Guinan, J. C. & Grehan, A. J. Multiscale terrain analysis of multibeam bathymetry data for habitat mapping on the continental slope. Mar. Geod. 30, 3–35. https://doi.org/10.1080/01490410701295962 (2007).Article 

Google Scholar 
raster: Geographic Data Analysis and Modelling. R package version 3.4–5 (2020).Shelton, A. O., Thorson, J. T., Ward, E. J., Feist, B. E. & Cooper, A. Spatial semiparametric models improve estimates of species abundance and distribution. Can. J. Fish. Aquat. Sci. 71, 1655–1666. https://doi.org/10.1139/cjfas-2013-0508 (2014).CAS 
Article 

Google Scholar 
Dunn, P. K. & Smyth, G. K. Series evaluation of Tweedie exponential dispersion model densities. Statiustics Comput. 15, 267–280 (2005).MathSciNet 
Article 

Google Scholar 
Gu, C. & Wahba, G. Minimizing GCV/GML scores with multiple smoothing parameters via the newton method. SIAM J. Sci. Stat. Comput. 12, 383–398. https://doi.org/10.1137/0912021 (1991).MathSciNet 
Article 
MATH 

Google Scholar 
Hobson, E. S. Feeding behaviour in three species of sharks. Pac. Sci. 17, 171–194 (1963).
Google Scholar 
Weeks, S., Magno-Canto, M., Jaine, F., Brodie, J. & Richardson, A. Unique sequence of events triggers manta ray feeding frenzy in the Southern Great Barrier Reef, Australia. Remote Sens. 7, 3138–3152. https://doi.org/10.3390/rs70303138 (2015).ADS 
Article 

Google Scholar 
Montero-Quintana, A. N., Ocampo-Valdez, C. F., Vázquez-Haikin, J. A., Sosa-Nishizaki, O. & Osorio-Beristain, M. Whale shark (Rhincodon typus) predatory flexible feeding behaviors on schooling fish. J. Ethol. 39, 399–410. https://doi.org/10.1007/s10164-021-00717-y (2021).Article 

Google Scholar 
Findlay, K. P. et al. Humpback whale “super-groups”—A novel low-latitude feeding behaviour of Southern Hemisphere humpback whales (Megaptera novaeangliae) in the Benguela Upwelling System. PLoS ONE 12, e0172002. https://doi.org/10.1371/journal.pone.0172002 (2017).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Pirotta, V., Owen, K., Donnelly, D., Brasier, M. J. & Harcourt, R. First evidence of bubble-net feeding and the formation of ‘super-groups’ by the east Australian population of humpback whales during their southward migration. Aquat. Conserv. Mar. Freshwat. Ecosyst. 31, 2412–2419. https://doi.org/10.1002/aqc.3621 (2021).Article 

Google Scholar 
Baines, M., Reichelt, M. & Griffin, D. An autumn aggregation of fin (Balaenoptera physalus) and blue whales (B. musculus) in the Porcupine Seabight, southwest of Ireland. Deep Sea Res. Part II Top. Stud. Oceanogr. 141, 168–177, https://doi.org/10.1016/j.dsr2.2017.03.007 (2017).Ladrón de Guevara P., P., Lavaniegos, B. E. & Heckel, G. Fin whales (Balaenoptera physalus) foraging on daytime surface swarms of the euphausiid Nyctiphanes simplex in Ballenas Channel, Gulf of California, Mexico. J. Mammal. 89, 559–566 (2008).Bruce, W. S. Some observations on Antarctic cetacea. In: Report on the Scientific results of the voyage of S.Y. ‘Scotia’ during the years 1902, 1903, and 1904, under the leadership of William S. Bruce, 491–505 (1915).Mackintosh, N. A. & Wheeler, J. F. G. Southern blue and fin whales. Discov. Rep. I, 257–540 (1929).Jackson, J. A. et al. Have whales returned to a historical hotspot of industrial whaling? The pattern of southern right whale Eubalaena australis recovery at South Georgia. Endang. Spec. Res. 43, 323–339. https://doi.org/10.3354/esr01072 (2020).Article 

Google Scholar 
Goldbogen, J. A., Pyenson, N. D. & Shadwick, R. E. Big gulps require high drag for fin whale lunge feeding. Mar. Ecol. Prog. Ser. 349, 289–301. https://doi.org/10.3354/meps07066 (2007).ADS 
Article 

Google Scholar 
Goldbogen, J. A. et al. Scaling of lunge-feeding performance in rorqual whales: mass-specific energy expenditure increases with body size and progressively limits diving capacity. Funct. Ecol. 26, 216–226. https://doi.org/10.1111/j.1365-2435.2011.01905.x (2012).Article 

Google Scholar 
Acevedo-Gutierrez, A., Croll, D. A. & Tershy, B. R. High feeding costs limit dive time in the largest whales. J Exp Biol 205, 1747–1753. https://doi.org/10.1242/jeb.205.12.1747 (2002).CAS 
Article 
PubMed 

Google Scholar 
Keen, E. M. Aggregative and feeding thresholds of sympatric rorqual whales within a fjord system. Ecosphere 8, e01702. https://doi.org/10.1002/ecs2.1702 (2017).Article 

Google Scholar 
Veit, R. R. & Harrison, N. M. Positive interactions among foraging seabirds, marine mammals and fishes and implications for their conservation. Front. Ecol. Evol. https://doi.org/10.3389/fevo.2017.00121 (2017).Article 

Google Scholar 
Laran, S. et al. A comprehensive survey of pelagic megafauna: their distribution, densities, and taxonomic richness in the tropical Southwest Indian Ocean. Front. Mar. Sci. https://doi.org/10.3389/fmars.2017.00139 (2017).Article 

Google Scholar 
Campbell, G. S. et al. Inter-annual and seasonal trends in cetacean distribution, density and abundance off southern California. Deep Sea Res. Part II Top. Stud. Oceanogr. 112, 143–157. https://doi.org/10.1016/j.dsr2.2014.10.008 (2015).ADS 
Article 

Google Scholar 
Heide-Jørgensen, M. P. et al. Estimates of large whale abundance in West Greenland waters from an aerial survey in 2005. J. Cetac. Res. Manag. 10, 119–129 (2008).
Google Scholar 
Panigada, S. et al. Estimating cetacean density and abundance in the Central and Western Mediterranean Sea through aerial surveys: Implications for management. Deep Sea Res. Part II Top. Stud. Oceanogr. 141, 41–58. https://doi.org/10.1016/j.dsr2.2017.04.018 (2017).ADS 
Article 

Google Scholar 
Forcada, J., Aguilar, A., Hammond, P., Pastor, X. & Aguilar, R. Distribution and abundance of fin whales (Balaenoptera physalus) in the western Mediterranean sea during the summer. J. Zool. 238, 23–34. https://doi.org/10.1111/j.1469-7998.1996.tb05377.x (2009).Article 

Google Scholar 
Clapham, P. J., Aguilar, A. & Hatch, L. T. Determining spatial and temporal scales for management: lessons from whaling. Mar. Mamm. Sci. 24, 183–201. https://doi.org/10.1111/j.1748-7692.2007.00175.x (2008).Article 

Google Scholar 
Baker, C. S. et al. Strong maternal fidelity and natal philopatry shape genetic structure in North Pacific humpback whales. Mar. Ecol. Prog. Ser. 494, 291–306. https://doi.org/10.3354/meps10508 (2013).ADS 
Article 

Google Scholar 
Carroll, E. L. et al. Cultural traditions across a migratory network shape the genetic structure of southern right whales around Australia and New Zealand. Sci. Rep. 5, 16182. https://doi.org/10.1038/srep16182 (2015).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Valenzuela, L. O., Sironi, M., Rowntree, V. J. & Seger, J. Isotopic and genetic evidence for culturally inherited site fidelity to feeding grounds in southern right whales (Eubalaena australis). Mol. Ecol. 18, 782–791. https://doi.org/10.1111/j.1365-294X.2008.04069.x (2009).CAS 
Article 
PubMed 

Google Scholar 
Barendse, J., Best, P. B., Carvalho, I. & Pomilla, C. Mother knows best: occurrence and associations of resighted humpback whales suggest maternally derived fidelity to a Southern Hemisphere coastal feeding ground. PLoS ONE 8, e81238. https://doi.org/10.1371/journal.pone.0081238 (2013).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Richard, G. et al. Cultural transmission of fine-scale fidelity to feeding sites may shape humpback whale genetic diversity in Russian Pacific waters. J. Hered. 109, 724–734. https://doi.org/10.1093/jhered/esy033 (2018).CAS 
Article 
PubMed 

Google Scholar 
Carroll, E. L. et al. Reestablishment of former wintering grounds by New Zealand southern right whales. Mar. Mamm. Sci. 30, 206–220. https://doi.org/10.1111/mms.12031 (2014).Article 

Google Scholar 
Calderan, S. V. et al. South Georgia blue whales five decades after the end of whaling. Endang. Spec. Res. 43, 359–373. https://doi.org/10.3354/esr01077 (2020).Article 

Google Scholar 
Santora, J. A. & Veit, R. R. Spatio-temporal persistence of top predator hotspots near the Antarctic Peninsula. Mar. Ecol. Prog. Ser. 487, 287–304. https://doi.org/10.3354/meps10350 (2013).ADS 
Article 

Google Scholar 
Scheidat, M. et al. Cetacean surveys in the Southern Ocean using icebreaker-supported helicopters. Polar Biol. 34, 1513–1522. https://doi.org/10.1007/s00300-011-1010-5 (2011).Article 

Google Scholar 
Williams, R., Hedley, S. L. & Hammond, P. S. Modeling distribution and abundance of Antarctic baleen whales using ships of opportunity. Ecol. Soc. 11, [online] http://www.ecologyandsociety.org/vol11/iss11/art11/ (2006).Secchi, E. et al. Encounter rates of whales around the Antarctic peninsula with special reference to humpback whales, Megaptera novaeangliae, in the Gerlache Strait 1997–98 to 1999–2000. Mem. Qld. Mus. 47, 571–578 (2001).
Google Scholar 
Thiele, D. et al. Seasonal variability in whale encounters in the Western Antarctic Peninsula. Deep Sea Res. Part II Top. Stud. Oceanogr. 51, 2311–2325. https://doi.org/10.1016/j.dsr2.2004.07.007 (2004).ADS 
Article 

Google Scholar 
Johnston, D. W., Friedlaender, A. S., Read, A. J. & Nowacek, D. P. Initial density estimates of humpback whales Megaptera novaeangliae in the inshore waters of the western Antarctic Peninsula during the late autumn. Endang. Spec. Res. 18, 63–71. https://doi.org/10.3354/esr00395 (2012).Article 

Google Scholar 
Zerbini, A. N. et al. Assessing the recovery of an Antarctic predator from historical exploitation. R Soc Open Sci 6, 190368. https://doi.org/10.1098/rsos.190368 (2019).ADS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Mackintosh, N. A. The distribution of southern blue and fin whales, in Whales, dolphins, and porpoises. (ed K. S. Norris) Ch. 8, 125–144 (University of California Press, 1966).Schmitt, N. et al. Mixed-stock analysis of humpback whales (Megaptera novaeangliae) on Antarctic feeding grounds. J. Cetac. Res. Manag. 14, 141–157 (2014).
Google Scholar 
Schall, E. et al. Humpback whale song recordings suggest common feeding ground occupation by multiple populations. Sci. Rep. 11, 18806. https://doi.org/10.1038/s41598-021-98295-z (2021).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Archer, F. I. et al. Revision of fin whale Balaenoptera physalus (Linnaeus, 1758) subspecies using genetics. J. Mammal. 100, 1653–1670. https://doi.org/10.1093/jmammal/gyz121 (2019).Article 

Google Scholar 
Archer, F. I. et al. Mitogenomic phylogenetics of fin whales (Balaenoptera physalus spp.): Genetic evidence for revision of subspecies. PLoS ONE 8, e63396. https://doi.org/10.1371/journal.pone.0063396 (2013).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Cabrera, A. A. et al. Fin whale (Balaenoptera physalus) mitogenomics: A cautionary tale of defining sub-species from mitochondrial sequence monophyly. Mol. Phylogenet. Evol. 135, 86–97. https://doi.org/10.1016/j.ympev.2019.02.003 (2019).Article 
PubMed 

Google Scholar 
Edwards, E. F., Hall, C., Moore, T. J., Sheredy, C. & Redfern, J. V. Global distribution of fin whales Balaenoptera physalus in the post-whaling era (1980–2012). Mammal. Rev. 45, 197–214. https://doi.org/10.1111/mam.12048 (2015).Article 

Google Scholar 
Knox, G. A. The key role of krill in the ecosystem of the Southern Ocean with special reference to the Convention on the Conservation of Antarctic Marine Living Resources. Ocean Manag. 9, 113–156 (1984).Article 

Google Scholar 
Ducklow, H. W. et al. Marine pelagic ecosystems: the west Antarctic Peninsula. Philos. Trans. R. Soc. Lond. B Biol. Sci. 362, 67–94. https://doi.org/10.1098/rstb.2006.1955 (2007).Article 
PubMed 

Google Scholar 
Atkinson, A. et al. Krill (Euphausia superba) distribution contracts southward during rapid regional warming. Nat. Clim. Change 9, 142–147. https://doi.org/10.1038/s41558-018-0370-z (2019).ADS 
Article 

Google Scholar 
Benton, M. J., Tverdokhlebov, V. P. & Surkov, M. V. Ecosystem remodelling among vertebrates at the Permian-Triassic boundary in Russia. Nature 432, 97–100. https://doi.org/10.1038/nature02950 (2004).ADS 
CAS 
Article 
PubMed 

Google Scholar 
Siegel, V., Reiss, C. S., Dietrich, K. S., Haraldsson, M. & Rohardt, G. Distribution and abundance of Antarctic krill (Euphausia superba) along the Antarctic Peninsula. Deep Sea Res. Part I Oceanogr. Res. Pap. 77, 63–74. https://doi.org/10.1016/j.dsr.2013.02.005 (2013).ADS 
Article 

Google Scholar 
Siegel, V. & Watkins, L. Distribution, Biomass and Demography of Antarctic Krill, Euphausia superba, in Biology and Ecology of Antarctic Krill Advances in Polar Ecology (ed Volker Siegel) Ch. 2, 21–101 (Springer, 2016).Roman, J. et al. Whales as marine ecosystem engineers. Front. Ecol. Environ. 12, 377–385. https://doi.org/10.1890/130220 (2014).Article 

Google Scholar 
Savoca, M. S. et al. Baleen whale prey consumption based on high-resolution foraging measurements. Nature 599, 85–90. https://doi.org/10.1038/s41586-021-03991-5 (2021).ADS 
CAS 
Article 
PubMed 

Google Scholar 
Atkinson, A., Siegel, V., Pakhomov, E. A., Jessopp, M. J. & Loeb, V. A re-appraisal of the total biomass and annual production of Antarctic krill. Deep Sea Res. Part I Oceangr. Res. Pap. 56, 727–740. https://doi.org/10.1016/j.dsr.2008.12.007 (2009).ADS 
Article 

Google Scholar 
Nicol, S. et al. Southern Ocean iron fertilization by baleen whales and Antarctic krill. Fish Fish. 11, 203–209. https://doi.org/10.1111/j.1467-2979.2010.00356.x (2010).Article 

Google Scholar 
Lavery, T. J. et al. Whales sustain fisheries: Blue whales stimulate primary production in the Southern Ocean. Mar. Mamm. Sci. 30, 888–904. https://doi.org/10.1111/mms.12108 (2014).CAS 
Article 

Google Scholar 
Smetacek, V. Are declining Antarctic krill stocks a result of global warming or of the decimation of the whales? in Effects of global warming on polar ecosystems (ed Carlos M. Duarte) (Fundación BBVA, 2008).Roman, J. & McCarthy, J. J. The whale pump: marine mammals enhance primary productivity in a coastal basin. PLoS ONE 5, e13255. https://doi.org/10.1371/journal.pone.0013255 (2010).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Smetacek, V. A whale of an appetite revealed by analysis of prey consumption. Nature 599, 33–34. https://doi.org/10.1038/d41586-021-02951-3 (2021).ADS 
CAS 
Article 
PubMed 

Google Scholar 
Laws, R. M. Seals and whales of the Southern Ocean. Philios. Trans. R. Soc. Lond. B 279, 81–96 (1977).ADS 
Article 

Google Scholar 
Lavery, T. J. et al. Iron defecation by sperm whales stimulates carbon export in the Southern Ocean. Proc. Biol. Sci. 277, 3527–3531. https://doi.org/10.1098/rspb.2010.0863 (2010).Article 
PubMed 
PubMed Central 

Google Scholar 
Cunningham, S. A. Transport and variability of the Antarctic Circumpolar Current in Drake Passage. J. Geophys. Res. https://doi.org/10.1029/2001jc001147 (2003).Article 

Google Scholar 
Frölicher, T. L. et al. Dominance of the Southern Ocean in Anthropogenic Carbon and Heat Uptake in CMIP5 Models. J. Clim. 28, 862–886. https://doi.org/10.1175/jcli-d-14-00117.1 (2015).ADS 
Article 

Google Scholar 
Landschutzer, P. et al. The reinvigoration of the Southern Ocean carbon sink. Science 349, 1221–1224. https://doi.org/10.1126/science.aab2620 (2015).ADS 
CAS 
Article 
PubMed 

Google Scholar  More