Philippa Kaur

Many eukaryotes display intraspecific genome size (GS) variation due to varying amounts of non-coding DNA1,2,3,4,5. Such GS variation can be mediated by additional genomic elements, which are physically represented either by extra (B-)chromosomes or by large heterozygous insertions into the regular chromosomes. On a DNA sequence level, non-coding DNA can be classified as highly repetitive, e.g. interspersedly repeated transposable elements or tandemly repeated satellite DNA, or as the result of previous duplications of the genome followed by pseudogenization6. The long-term gain and loss of such non-coding DNA sequences is thought to be governed by largely neutral evolutionary processes, and their excessive accumulation in some genomes can be explained by genetic drift7,8, even though selection might also sometimes play a role9,10.Non-coding DNA can affect organisms in different ways. A large number of studies document correlations between genome size and organismic traits such as cell size11,12, body size13,14, or developmental rates15, sometimes even at the within-population level13. Under some circumstances, differential amounts of non-coding DNA might even affect fitness16. Furthermore, DNA can have coding-independent effects that operate at lower levels, such as intragenomic selection. For example, (additional) genomic elements might increase their own fitness by increasing their transmission rates to offspring by meiotic drive, sometimes at the expense of their host’s fitness17,18,19. Meiotic drive in this classical sense occurs during the chromosome segregation during the meiotic divisions, even though later stages during gametogenesis can also be affected20. Recognizing and disentangling such effects is important for a better understanding of the evolution of eukaryotic genomes, in particular, the evolutionary causes of the large intraspecific genome size variation.Here we study meiotic transmission patterns of additional genomic elements in the monogonont rotifer Brachionus aplanchnoidis. Individuals of this species can differ by up to almost two-fold in genome size, which is mediated by several Megabase-sized independently segregating genomic elements (ISEs) consisting mainly of tandemly repeated satellite DNA21. The genomic data are consistent with a mixture of both B-chromosomes and large insertions to normal chromosomes21,22. Individual rotifers and their clonal offspring can be characterized by the number and size of their ISEs and their composition stays constant through hundreds of asexual (mitotic) generations22. Occasionally, monogonont rotifers engage in sexual reproduction (Fig. 1), producing sexual females, whose oocytes undergo classical meiosis with two polar bodies formed23. Unfertilized haploid eggs develop mitotically into males, and sperm production does not involve any meiotic maturation divisions24. By analyzing the genome size distributions of haploid males produced by different mother clones, it has been shown that ISEs segregate in a manner suggesting that they do not pair with each other, nor with any other part of the genome22. For instance, a clone containing three ISEs will produce males (and gametes) that might contain either zero, one, two, or three ISEs, corresponding to four different GS classes of the males in this clone. The frequencies of these different GS classes roughly approximated those expected by random segregation. However, previous studies in B. asplanchnoidis did not resolve different steps during meiotic transmission, so they were not designed to detect meiotic drive or subsequent changes in meiotic transmission, and they also did not test whether there were subtle deviations from completely independent segregation.Figure 1Schematics of rotifer life cycle. Monogonont rotifers are cyclical parthenogens, capable of both ameiotic parthenogenesis and sexual reproduction. The production of sexual females is triggered by quorum sensing chemicals, released by the animals themselves at high population density. In contrast to parthenogenetic females, sexual females produce oocytes by meiosis, and give rise to either haploid males or diploid resting eggs, depending on whether they get fertilized by a male24.Full size imageIn the present study, we test for meiotic transmission biases of ISEs. If meiotic transmission would be completely unbiased, the frequencies of haploid oocytes, or males, with different numbers of ISEs should be identical to those expected by random segregation. For example, a mother with two ISEs should produce males with zero, one, or two ISEs (hence, three male GS classes), which have relative frequencies of 0.25, 0.5, and 0.25, respectively. However, if ISEs avoid segregating into polar bodies due to meiotic drive17,20,25, one would expect to see an increase in the relative frequency of male GS classes with two ISEs, compared to those with no ISE . By contrast, if ISEs are preferentially sequestered into polar bodies due to meiotic drag 7,26, the GS class with two ISEs should be underrepresented. Our experimental approach for detecting meiotic transmission biases relies on measuring (by flow-cytometry) the observed relative frequencies of each male GS class and comparing these to their relative frequencies expected under unbiased transmission (Fig. 2). To allow for clear comparisons, the main output variable in these analyses is the observed/expected ratio (O/E-ratio), i.e., the observed frequency divided by the expected relative frequency for each GS class. If there were no transmission biases, O/E-ratios across all GS classes should equal one. In contrast, O/E-ratios larger than one indicate overrepresentation of a certain GS class, and if O/E ratios increase or decrease with genome size, this indicates drive or drag at a meiotic or postmeiotic stage (Fig. 2d,h).Figure 2Principle of inferring meiotic transmission patterns from the genome size distributions of haploid rotifer males. The first four panels (a–d) show a rotifer clone with one ISE (i.e., two corresponding male GS classes). The last four panels (e–h) show a clone with four ISEs (i.e., five corresponding male GS classes). a, e Example of flow cytometry data. b, f Conceptual model of ISE meiotic segregation. c, g Theoretically predicted GS distributions of males (relative to the female GS) under meiotic drive, meiotic drag, or in the absence of meiotic drive. d, h Theoretically predicted O/E ratios (observed vs. expected frequencies of different male GS classes) under drive, drag, or on absence of drive. O/E values of  > 1 indicate over-representation of a GS class (relative to the frequency expected from unbiased transmission).Full size imageWe implemented these ideas in a mathematical model that contains the two parameters, transmission bias and cosegregation bias. Values for transmission bias may range from − 1 to 1 in our model. For instance, a value of 0.1 denotes a 10% increase in probability that an ISE segregates towards the egg pole (this is equivalent to a transmission rate of 0.55 for this ISE, i.e. mild meiotic drive). Concerning the second parameter, cosegregation bias, a positive value means that pairs of ISEs have an increased probability of being sequestered towards the same pole (irrespective of whether this is the egg pole or polar body pole), while a negative bias favors migration towards opposite poles. Please note that a cosegregation bias value of − 1 (i.e., 100% probability that ISEs migrate towards opposite poles) resembles the default segregation pattern of regular chromosomes. By estimating the transmission bias and cosegregation bias parameter for each rotifer clone, we tried to infer and compare general meiotic transmission patterns across clones, even if they contained different numbers and types of ISEs.Transmission biases may not only arise during meiosis, as described above but also during later stages of male embryonic development. For instance, they might be caused by differences in the survival of embryos, or due to differences in the fitness of hatched males containing different numbers of ISEs. To address these potential sources of variation, we compared the transmission biases in relatively young, synchronized male eggs, older eggs accumulating in growing cultures, and hatched males. Finally, to address the question of whether a high number of ISEs affects male embryonic survival in general, we estimated and compared hatching rates of (haploid) male eggs and (diploid) female eggs in 19 rotifer clones of different genome sizes (which is highly correlated with the number and size of ISEs in the genome22).Our results suggested that the ISEs in B. asplanchnoidis exhibit diverse meiotic segregation patterns: In some rotifer clones, transmission bias was positive, while the ISEs of other clones showed negative transmission bias (indicative of drag). Furthermore, we obtained evidence for a negative cosegregation bias in some clones, i.e., pairs of ISEs showed an increased probability to segregate towards opposite poles. Overall, these transmission patterns seemed to be determined early in the haploid life cycle, probably at or shortly after meiosis, since early and late stages of male embryonic development showed very similar GS distributions. Finally, we found that very large genome size (i.e., a large numbers of ISEs) was associated with reduced male embryonic survival. More

United Nations. World Urbanization Prospects: The 2018 revision. (Department of Economic and Social Affairs, Population Division, United Nations, 2018).Grimm, N. B. et al. Global change and the ecology of cities. Science 319, 756–760 (2008).Article 

Google Scholar 
McPhearson, T. et al. Advancing urban ecology toward a science of cities. Bioscience 66, 198–212 (2016).Article 

Google Scholar 
Dodman, D. et al. Cities, settlements and key infrastructure. In Climate Change 2022: Impacts, Adaptation, and Vulnerability. Contribution of Working Group II to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change (eds. Pörtner, H.-O. et al.) 997–1040 (Cambridge University Press, Cambridge, UK and New York, NY, USA, 2022).Díaz, S. et al. Assessing nature’s contributions to people: Recognizing culture, and diverse sources of knowledge, can improve assessments. Science 359, 270–272 (2018).Article 

Google Scholar 
Grabowski, Z. J., McPhearson, T., Matsler, A. M., Groffman, P. & Pickett, S. T. A. What is green infrastructure? A study of definitions in US city planning. Front. Ecol. Environ. 20, 152–160 (2022).Article 

Google Scholar 
Childers, D. L. et al. Urban ecological infrastructure: An inclusive concept for the non-built urban environment. Elementa 7, 1–14 (2019).
Google Scholar 
Gómez-Baggethun, E. et al. Urban ecosystem services. In Urbanization, Biodiversity and Ecosystem Services: Challenges and Opportunities (eds. Elmqvist, T. et al.) 175–251 (Springer, Netherlands, 2013).Díaz, S. & Cabido, M. Vive la différence: Plant functional diversity matters to ecosystem processes. Trends Ecol. Evol. 16, 646–655 (2001).Article 

Google Scholar 
Burkhard, B. & Maes, J. Mapping Ecosystem Services (Pensoft Publishers, Sofia, 2017).Eviner, V. T. & Chapin, F. S. Functional Matrix: A conceptual framework for predicting multiple plant effects on ecosystem processes. Annu. Rev. Ecol. Evol. Syst. 34, 455–485 (2003).Article 

Google Scholar 
Lavorel, S., McIntyre, S., Landsberg, J. & Forbes, T. D. A. Plant functional classifications: From general groups to specific groups based on response to disturbance. Trends Ecol. Evol. 12, 474–478 (1997).Article 

Google Scholar 
Cornelissen, J. H. C. et al. A handbook of protocols for standardised and easy measurement of plant functional traits worldwide. Aust. J. Bot. 51, 335–380 (2003).Article 

Google Scholar 
Suding, K. N. et al. Scaling environmental change through the community-level: A trait-based response-and-effect framework for plants. Glob. Chang. Biol. 14, 1125–1140 (2008).Article 

Google Scholar 
Lavorel, S. & Garnier, E. Predicting changes in community composition and ecosystem functioning from plant traits: Revisiting the Holy Grail. Funct. Ecol. 16, 545–556 (2002).Article 

Google Scholar 
Hevia, V. et al. Trait-based approaches to analyze links between the drivers of change and ecosystem services: Synthesizing existing evidence and future challenges. Ecol. Evol. 7, 831–844 (2017).Article 

Google Scholar 
Cadotte, M. W., Carscadden, K. & Mirotchnick, N. Beyond species: Functional diversity and the maintenance of ecological processes and services. J. Appl. Ecol. 48, 1079–1087 (2011).Article 

Google Scholar 
Lavorel, S. Plant functional effects on ecosystem services. J. Ecol. 101, 4–8 (2013).Article 

Google Scholar 
Andersson, E. et al. What are the traits of a social-ecological system: Towards a framework in support of urban sustainability. npj Urban Sustain. 1, 14 (2021).Article 

Google Scholar 
Pickett, S. T. A. et al. Urban ecological systems: Linking terrestrial ecological, physical, and socioeconomic components of metropolitan areas. Annu. Rev. Ecol. Syst. 32, 127–157 (2001).Article 

Google Scholar 
McPhearson, T., Haase, D., Kabisch, N. & Gren, Å. Advancing understanding of the complex nature of urban systems. Ecol. Indic. 70, 566–573 (2016).Article 

Google Scholar 
Zhou, W., Pickett, S. T. A. & McPhearson, T. Conceptual frameworks facilitate integration for transdisciplinary urban science. npj Urban Sustain. 1, 1 (2021).Article 

Google Scholar 
Andersson, E. et al. Scale and context dependence of ecosystem service providing units. Ecosyst. Serv. 12, 157–164 (2015).Article 

Google Scholar 
Pinho, P. et al. Research agenda on biodiversity and ecosystem functions and services in European cities. Basic Appl. Ecol. 53, 124–133 (2021).Article 

Google Scholar 
Bullock, J. M. et al. Human-mediated dispersal and the rewiring of spatial networks. Trends Ecol. Evol. 33, 958–970 (2018).Article 

Google Scholar 
Avolio, M. L., Swan, C., Pataki, D. E. & Jenerette, G. D. Incorporating human behaviors into theories of urban community assembly and species coexistence. Oikos 130, 1849–1864 (2021).Article 

Google Scholar 
Aronson, M. F. J. et al. Hierarchical filters determine community assembly of urban species pools. Ecology 97, 2952–2963 (2016).Article 

Google Scholar 
Woodward, F. I. & Diament, A. D. Functional approaches to predicting the ecological effects of global change. Funct. Ecol. 5, 212 (1991).Article 

Google Scholar 
Diaz, S., Cabido, M. & Casanoves, F. Plant functional traits and environmental filters at a regional scale. J. Veg. Sci. 9, 113–122 (1998).Article 

Google Scholar 
Boet, O., Arnan, X. & Retana, J. The role of environmental vs. biotic filtering in the structure of European ant communities: A matter of trait type and spatial scale. PLoS ONE 15, e0228625 (2020).Article 

Google Scholar 
Grimm, N. B., Grove, J. M., Pickett, S. T. A. & Redman, C. L. Integrated approaches to long-term studies of urban ecological systems. Bioscience 50, 571–584 (2000).Article 

Google Scholar 
Vandewalle, M. et al. Functional traits as indicators of biodiversity response to land use changes across ecosystems and organisms. Biodivers. Conserv. 19, 2921–2947 (2010).Article 

Google Scholar 
Williams, N. S. G. et al. A conceptual framework for predicting the effects of urban environments on floras. J. Ecol. 97, 4–9 (2009).Article 

Google Scholar 
Cavender-Bares, J. et al. Horticultural availability and homeowner preferences drive plant diversity and composition in urban yards. Ecol. Appl. 30, 1–16 (2020).Article 

Google Scholar 
Pearse, W. D. et al. Homogenization of plant diversity, composition, and structure in North American urban yards. Ecosphere 9, e02105 (2018).Article 

Google Scholar 
Cubino, J. P. et al. Drivers of plant species richness and phylogenetic composition in urban yards at the continental scale. Landsc. Ecol. 34, 63–77 (2019).Article 

Google Scholar 
Oke, T. R. The energetic basis of the urban heat island. Q. J. R. Meteorol. Soc. 108, 1–24 (1982).
Google Scholar 
Sukopp, H. Human-caused impact on preserved vegetation. Landsc. Urban Plan. 68, 347–355 (2004).Article 

Google Scholar 
Díaz, S. et al. Incorporating plant functional diversity effects in ecosystem service assessments. Proc. Natl. Acad. Sci. USA 104, 20684–20689 (2007).Article 

Google Scholar 
Williams, N. S. G., Hahs, A. K. & Vesk, P. A. Urbanisation, plant traits and the composition of urban floras. Perspect. Plant Ecol. Evol. Syst. 17, 78–86 (2015).Article 

Google Scholar 
Teskey, R. et al. Responses of tree species to heat waves and extreme heat events. Plant Cell Environ. 38, 1699–1712 (2015).Article 

Google Scholar 
Jochner, S. & Menzel, A. Urban phenological studies—past, present, future. Environ. Pollut. 203, 250–261 (2015).Article 

Google Scholar 
Cleland, E. E., Chuine, I., Menzel, A., Mooney, H. A. & Schwartz, M. D. Shifting plant phenology in response to global change. Trends Ecol. Evol. 22, 357–365 (2007).Article 

Google Scholar 
de Bello, F. et al. Towards an assessment of multiple ecosystem processes and services via functional traits. Biodivers. Conserv. 19, 2873–2893 (2010).Article 

Google Scholar 
Santangelo, J. S. et al. Global urban environmental change drives adaptation in white clover. Science 375, 1275–1281 (2022).Article 

Google Scholar 
Blois, J. L., Zarnetske, P. L., Fitzpatrick, M. C. & Finnegan, S. Climate change and the past, present, and future of biotic interactions. Science 341, 499–504 (2013).Article 

Google Scholar 
Martin, C. A., Warren, P. S. & Kinzig, A. P. Neighborhood socioeconomic status is a useful predictor of perennial landscape vegetation in residential neighborhoods and embedded small parks of Phoenix, AZ. Landsc. Urban Plan. 69, 355–368 (2004).Article 

Google Scholar 
Kinzig, A. P., Warren, P., Martin, C., Hope, D. & Katti, M. The effects of human socioeconomic status and cultural characteristics on urban patterns of biodiversity. Ecol. Soc. 10, 23 (2005).Article 

Google Scholar 
Stephenson, J. The cultural values model: An integrated approach to values in landscapes. Landsc. Urban Plan. 84, 127–139 (2008).Article 

Google Scholar 
Andersson, E., Barthel, S. & Ahrné, K. Measuring social–ecological dynamics behind the generation of ecosystem services. Ecol. Appl. 17, 1267–1278 (2007).Article 

Google Scholar 
Fraser, E. D. G. & Kenney, W. A. Cultural background and landscape history as factors affecting perceptions of the urban forest. J. Arboric. 26, 106–113 (2000).
Google Scholar 
Hope, D. et al. Socioeconomics drive urban plant diversity. Proc. Natl. Acad. Sci. USA 100, 8788–8792 (2003).Article 

Google Scholar 
Avolio, M. L. et al. Understanding preferences for tree attributes: The relative effects of socio-economic and local environmental factors. Urban Ecosyst. 18, 73–86 (2015).Article 

Google Scholar 
Körmöndi, B., Tempfli, J., Kocsis, J. B., Adams, J. & Szkordilisz, F. E. The secret ingredient—The role of governance in green infrastructure development: Through the examples of European cities. IOP Conf. Ser. Earth Environ. Sci. 323, (2019).Conway, T. M. & Vander Vecht, J. Growing a diverse urban forest: species selection decisions by practitioners planting and supplying trees. Landsc. Urban Plan. 138, 1–10 (2015).Article 

Google Scholar 
Lack, W. H. The Book of Palms (Taschen-Bibliotheca Universalis, 2015).Grilo, F. et al. Using green to cool the grey: Modelling the cooling effect of green spaces with a high spatial resolution. Sci. Total Environ. 724, 138182 (2020).Article 

Google Scholar 
Prasifka, J. R. et al. Using nectar-related traits to enhance crop–pollinator interactions. Front. Plant Sci. 9, 1–8 (2018).Article 

Google Scholar 
Veerkamp, C. J. et al. A review of studies assessing ecosystem services provided by urban green and blue infrastructure. Ecosyst. Serv. 52, 101367 (2021).Article 

Google Scholar 
Theodorou, P. et al. Urban areas as hotspots for bees and pollination but not a panacea for all insects. Nat. Commun. 11, 576 (2020).Article 

Google Scholar 
Farmer, J. Trees in Paradise: A California History (WW Norton & Company, 2013).Goodness, J., Andersson, E., Anderson, P. M. L. & Elmqvist, T. Exploring the links between functional traits and cultural ecosystem services to enhance urban ecosystem management. Ecol. Indic. 70, 597–605 (2016).Article 

Google Scholar 
Masterson, V. A. et al. The contribution of sense of place to social-ecological systems research: A review and research agenda. Ecol. Soc. 22, 49 (2017).Article 

Google Scholar 
Masterson, V. A., Enqvist, J. P., Stedman, R. C. & Tengö, M. Sense of place in social-ecological systems: From theory to empirics. Sustain. Sci. 14, 555–564 (2019).Article 

Google Scholar 
Mukherjee, A. & Agrawal, M. Use of GLM approach to assess the responses of tropical trees to urban air pollution in relation to leaf functional traits and tree characteristics. Ecotoxicol. Environ. Saf. 152, 42–54 (2018).Article 

Google Scholar 
Singh, S. K., Rao, D. N., Agrawal, M., Pandey, J. & Naryan, D. Air pollution tolerance index of plants. J. Environ. Manage. 32, 45–55 (1991).Article 

Google Scholar 
Mukherjee, A. & Agrawal, M. Pollution response score of tree species in relation to ambient air quality in an urban area. Bull. Environ. Contam. Toxicol. 96, 197–202 (2016).Article 

Google Scholar 
Barwise, Y. & Kumar, P. Designing vegetation barriers for urban air pollution abatement: A practical review for appropriate plant species selection. npj Clim. Atmos. Sci. 3, 12 (2020).Article 

Google Scholar 
Grote, R. et al. Functional traits of urban trees: Air pollution mitigation potential. Front. Ecol. Environ. 14, 543–550 (2016).Article 

Google Scholar 
Tomson, M. et al. Green infrastructure for air quality improvement in street canyons. Environ. Int. 146, 106288 (2021).Article 

Google Scholar  More

Knowing the distribution of species is crucial for effective conservation action. However, accurate and high-resolution spatial data are only available for a limited number of species1,2. For mammals and birds, the most comprehensive and widely used global distribution dataset is the set of range maps compiled as part of the assessments for the International Union for Conservation of Nature (IUCN) Red List. These represent each species’ distributional limits and tend to minimize omission errors (i.e. false absences) at the expense of commission errors (i.e. false presences)3,4. Therefore, they often contain sizeable areas not regularly occupied by the species.Maps of the Area of Habitat (AOH; previously known as Extent of Suitable Habitat, ESH) complement range maps by indicating potential occupancy within the range, thereby reducing commission errors5. AOH is defined as ‘the habitat available to a species, that is, habitat within its range’5. These models are produced by subtracting areas unsuitable for the species within their range, using information on each species’ associations with habitat and elevation5,6,7,8. Comprehensive sets of AOH maps have been produced in the past for mammals6 and amphibians7, as well as subsets of birds8,9. The percentage of a species’ range covered by the AOH varies depending on the methodology used to associate species to their habitats, and their habitats to land-cover, the coarseness of the range map, the region in which the species is distributed, and the species’ habitat specialization and elevation limits5. For example, Rondinini et al.6 found that, when considering elevation and land cover features for terrestrial mammals, the AOH comprised, on average, 55% of the range. Ficetola et al.7 obtained a similar percentage when analyzing amphibians (55% for forest species, 42% for open habitat species and 61% for habitat generalists). Beresford et al.8 found that AOH covered a mean of 27.6% of the range maps of 157 threatened African bird species. In 2019, Brooks et al.5 proposed a formal definition and standardized methodology to produce AOH, limiting the inputs to habitat preferences, elevation limits, and geographical range.AOH production requires knowledge of which habitat types a species occurs in and their location within the range1. Information on habitat preference is documented for each species assessed in the IUCN Red List10, following the IUCN Habitats Classification Scheme11. However, the IUCN does not define habitat classes in a spatially explicit way, therefore, we used a recently published translation table that associates IUCN Habitat Classification Scheme classes with land cover classes12. Species’ elevation limits were also extracted from the IUCN Red List.We developed AOH maps for 5,481 terrestrial mammal species and 10,651 terrestrial bird species (Fig. 1). For 1,816 bird species defined by BirdLife International as migratory, we developed separate AOH maps, for the resident, breeding, and non-breeding ranges, according to the migratory distribution of the species (Fig. 2). The maps are presented in a regular latitude/longitude grid with an approximate 100 m resolution at the equator. On average, the AOH covers 66 ± 28% of the geographical range for mammals and 64 ± 27% for birds. We used the resulting AOH maps to produce four global species richness layers for: mammals, birds, globally threatened mammals and globally threatened birds13 (Fig. 3).Fig. 1Spatial distribution maps of Tangara abbas. Maps represent (a) the geographic range21, and (b) the Area of Habitat (AOH) of the species. The AOH was produced by subtracting unsuitable habitats from the geographical range. This species’ habitats are forest and terrestrial artificial habitats and has elevation range of 0 – 1600 m.Full size imageFig. 2Spatial distribution maps of Cardellina rubrifrons, divided into resident, breeding and non-breeding areas for this migratory species. Maps represent (a) the geographic range21, and (b) the Area of Habitat (AOH) of the species. The AOH was produced by subtracting unsuitable habitats from the ranges. This species is a forest species with elevation rangelimits of 1500 – 3100 m.Full size imageFig. 3Global species richness maps for (a) terrestrial mammals (considering 5,481species) and (b) terrestrial birds (considering 10,651 species). Calculated by overlaying all species’ AOH per class, resulting inon the number of species at each grid cell, latitude/longitude grid at a resolution of 1°/1008 or approximately 100 m at the equator (EPSG:4326) with the ellipsoid WGS 1984.Full size imageThe AOH maps presented in this paper are more useful for some purposes than global species distribution models, as they reduce and standardize commissions14. They are especially useful for not well-known and wide-range species. However, we note that for well-known species alternative sources may have more accurate distributions15. Moreover, AOHs are affected by the bias and errors of the underlying data, especially relevant errors associated with documentation of species’ habitats and elevations, and the translation of habitats into land cover classes, given that habitat is a complex multidimensional concept that is challenging to match to land-cover classes12, and that the current version of the IUCN Habitat Classification Scheme on IUCN’s website is described as a draft version11.The AOH maps have multiple conservation applications5,16,17, such as assessing species’ distributions and extinction risk, improving the accuracy of conservation planning, monitoring habitat loss and fragmentation, and guiding conservation actions. AOH has been proposed as an additional spatial metric to be documented in the Red List5, and is used for the identification of Key Biodiversity Areas18. More

Study sites and study designThe study was performed in the frame of a dog ecology research project, with details on the study locations published elsewhere15,42,43. For the current study, five study sites located in Indonesia and Guatemala were included. Site selection was carried out by each country’s research team, taking into consideration rural and urban settings, as well as differing expected number of dogs present at each location. The Indonesian study sites were semi-urban Habi and rural Pogon, in the Sikka regency, at the eastern area of Flores Island (Supplementary Fig. 6). In Guatemala, the study sites were Poptún (urban setting), Sabaneta and La Romana (both rural settings), located in the Guatemalan department of Péten, in the northern part of the country (Supplementary Fig. 7). Data were collected during May to June 2018 in Guatemala and from July to September 2018 in Indonesia.In each location, a 1 km2 area was predefined using Google Earth within which the study took place. The 1 km2 area was chosen because of the research goals of another part of the project, investigating the contact network of the dogs15. Within these areas, the teams visited all dog-owning households. In each household, the study was presented to an adult of the family, who was then asked if they owned a dog and if they were willing to participate in the study. After the dog owner’s oral or written consent was granted, a questionnaire was answered, and the dogs collared. The handling of the dogs was performed by a trained veterinarian or a trained veterinary paramedic of the team.The questionnaire data was collected through interviews with the dog owners. Multiple dogs per household could be included as multiple entries in the questionnaire. The detailed questionnaire contains information on the household location, dog demographics (age, sex, reproductive status) and management (dog’s purpose, origin, confinement, vaccination status, feeding and human-mediated transportation within and outside the pre-determined area).All dogs of a household fulfilling the inclusion criteria were equipped with a geo-referenced contact sensor (GCS) developed by Bonsai Systems (https://www.bonsai-systems.com), containing a GPS module and an Ultra-High-Frequency (UHF) sensor for contact data recording43,44. GCS devices report a 5-m maximum accuracy, a run-time of up to 10 years, can store up to 4 million data points and carry a lithium-polymer-battery (LiPo). For this study, only GPS data were analysed. The GCS were set to record each dog’s geographical position at one-minute intervals. Dogs remained collared for 3 to 5 days with the duration of the data collection being limited by the device’s battery capacity, as batteries were not re-charged or changed during the study. Throughout the time of recording, date, hour, GPS coordinates and signal quality (HDOP) raw data were collected by the GPS module and amassed into the workable databases.Exclusion criteria were dogs of less than four months of age (since they were not big enough to carry a collar), sick dogs and pregnant bitches (to avoid any risk of stress-induced miscarriages). Reasons for non-participation of eligible dogs included dog owner’s absence, dog’s absence, inability to catch the dog, and refusal of participation by the dog owner. In addition, dogs foreseen for slaughtering within the following four days were excluded in Indonesia to ensure data collection for at least four to five days. All dogs included in this study were constantly free roaming or at least part-time (day only, night only and for some hours a day). Human and/or animal ethical approval were obtained depending on the country-specific regulations. All the procedures were carried out in accordance with relevant guidelines. Ethical clearance was granted in Guatemala by the UVG’s International Animal Care and Use Committee [Protocol No. I-2018(3)] and the Community Development Councils of the two rural sites, which included Maya Q’eqchi’ communities45. In Indonesia, the study was approved by the Animal Ethics Commission of the Faculty of Veterinary Medicine, Nusa Cendana University (Protocol KEH/FKH/NPEH/2019/009). In addition, dogs that participated in the study were vaccinated against rabies and/or dewormed to acknowledge the owners for their participation in the study.Data cleaningData were stored in an application developed by Bonsai Systems compatible with Apple operating system (iOS iPhone Operating Systems), downloaded as individual csv file for each unit, and further analysed in R (version 3.6.1)46.The GPS data were cleaned based on three automatised criteria. First, the speed was calculated between any two consecutive GPS fixes, and fixes with speed of  > 20 km/h were excluded, given the implausibility of a dog running at such speed over a one-minute timespan47. It is noteworthy that car travel causes speeds over 20 km/h. However, as we were interested in analysing the dog’s behaviour outside of car transports, removing these fixes was in line with our objectives. Second, the Horizontal Dilution of Precision (HDOP), which is a measure of accuracy48 and automatically recorded by the devices for each GPS fix, was used to exclude fixes with low precision. According to Lewis et al.49, GPS fixes with HDOP higher than five were excluded, which deleted 1.3% of data in Habi, 2.2% in Pogon, 3.3% in Poptún, 1.8% in La Romana and 2.1% in Sabaneta. Third, the angles built by three consecutive fixes were calculated for each dog. When studying animals’ trajectories as their measure of movement, acute inner angles are often connected to error GPS fixes50. The fixes having the 2.5% smallest angles were excluded, to target those fixes with highest risks of being errors, while balancing against the loss of GPS fixes due to the cleaning process. With the exclusion of the smallest angles, 2.6% of data were deleted in Habi, 3% in Pogon, 2.9% in Poptún, 2.6% in La Romana and 2.7% in Sabaneta. After the automatised cleaning was concluded, 18 obvious error GPS fixes (unachievable or inexplicable locations by dogs) still prevailed in the Habi dataset and were manually removed.Habitat resource identification and calculation of terrain slopeTo analyse habitat selection of the collared FRDD, resources were delimited by a 100% Minimum Convex Polygon (MCP) including all cleaned GPS fixes per study site, using QGIS51 (Fig. 1).Figure 1GPS fixes plotted over a Google satellite imagery layer with its respective outlined computed Minimum Convex Polygon (MCP) delimitating the habitat available for the study population in: (a) Habi; (b) Pogon; (c) Poptún; (d) La Romana and (e) Sabaneta. Source QGIS (version 3.4 Madeira, http://qgis.org), map data: Google Satellite.Full size imageResources were defined by taking into consideration the following criteria: resources are (i) likely to impact upon movement patterns of dogs, (ii) identifiable by landscape satellite topography, and (iii) chosen considering information on relevant gathering places for FRDD observed by the field teams. Three resources were disclosed in all study sites: buildings, roads and vegetation coverage. All habitat relevant resources were manually identified within the available area (MCP) in QGIS using satellite imagery. All building-like structures were identified using vector polygons and summed under the layer “buildings”. Roads were identified and manually traced using vector lines in all sites, except in Poptún where the roads were automatically traced using an OpenStreetMap road layer of the area (https://www.openstreetmap.org/export). A buffer vector polygon was generated to encompass the full potential width of the roads, with a 5 m width in Habi and Poptún (semi-urban and urban site) and a 2 m width in Pogon, La Romana and Sabaneta (rural sites). In Habi, a “beach” layer was defined by generating a five-meter buffer from the shoreline in both directions using a vector polygon. The layer “sea” was defined as the vector polygon resulting from the difference between the MCP sea outer limit and the beach buffer polygon. Vegetation coverage was distinct between study sites with sparse vegetation and bushes present in all sites except Pogon, and dense forest-like vegetation present in La Romana and Pogon. These two types of vegetation were defined as “low” and “high vegetation”, respectively. In Habi and La Romana, “low” and “high vegetation”, respectively, were manually identified using vector polygons and summarised under the respective layers. Finally, open field in Habi, high vegetation in Pogon and low vegetation in Poptún, La Romana and Sabaneta were the last vector layers to be established since they represented the difference between all other polygon vector layers and the MCP total area. After all resource vector polygons had been created, an encompassing vector layer was generated by merging all resource polygon vectors for final resource classification (Fig. 2). As part of the resource classification in Habi, the airport terminal and runaway as well as waterways enclosed in the MCP area were identified but excluded from the analysis.Figure 2(a) Habi, (b) Pogon, (c) Poptún, (d) La Romana and (e) Sabaneta Habitat classification vector layers. The different habitat resources, identifiable by colour, were merged to create the comprehensive Habitat classification vector. In the Indonesian sites (a, b) and Guatemalan sites (c–e) buildings are coloured red, vegetation low in Habi, Poptún, La Romana and Sabaneta is coloured light green, vegetation high in Pogon and La Romana dark green, roads black, beach yellow, sea dark blue, airport grey, waterways light blue and open field light orange. The airport area (gray) and waterways (light blue) in Habi were not classified as separate habitat layers and were excluded from further analysis. Source QGIS (version 3.4 Madeira, http://qgis.org), map data: Google Satellite.Full size imageAfter the construction of the habitat resource layers, all GPS fixes were assigned to the respective resource they were located, using the QGIS join attributes by location algorithm. Fixes located exactly on the MCP border in Indonesia were not classified automatically and had to be manually classified to the respective resource.In non-flat topographies (all locations expect Habi) we tested the hypothesis of whether the steepness would influence the dogs’ movement patterns. The degrees of slope were calculated using a 30-m raster-cell resolution (STRM 1-Arc Second Global, downloaded from the United States Geological Survey (USGS) Earth Explorer, https://earthexplorer.usgs.gov/). The slope was assigned by the QGIS join attributes by location algorithm to each GPS fix.Statistical analysisTo quantify habitat selection in each study site, we compared resources used by the dogs with the resources available, according to Freitas et al.52. Adapting the methodology applied by O’Neill et al.18, the observed number of GPS fixes for each dog was used to generate an equivalent number of locations that were randomly distributed within the MCP area using the Random points in layer bound vector tool from QGIS. For example, if dog “D300” had 100 recorded GPS fixes, 100 random points were generated within the MCP of the respective study site and assigned to “D300”. Random points were then assigned to the respective resources and slope of that location, as previously done with the observed GPS fixes. Using this approach, the habitat resources used by each dog could be compared to the available resources in the respective study site, using a regression model.Observation independence is a fundamental presupposition of any regression model. However, the spatial nature of the point-referenced data permits perception of spatial dependence. In our dataset, spatial autocorrelation was proven for all study sites using the Moran’s I test. Therefore, we applied a spatial regression model, which takes into consideration spatial autocorrelation while exploring the effects of the study variables. A mixed effects logistic regression model accounting for spatial autocorrelation was created to quantify the effect of variables on used (i.e. observed GPS fix) versus available (i.e. randomly generated GPS fixes) resources, using the fitme function in the spaMM package in R53,54. The model’s binary outcome variable was defined as either observed (1) or random (0) GPS fix, i.e. the dog being present or absent from a position. The explanatory variable was the resource classification with “buildings”, “roads”, “low vegetation”, “beach”, “sea” and “open field” as levels in Habi; “buildings”, “roads” and “high vegetation” in Pogon; “buildings”, “roads”, “low vegetation” in Poptún and Sabaneta; and “buildings”, “roads”, and “high” and “low vegetation” in La Romana. Different habitat resources were used interchangeably as reference level. In all study sites except Habi, the slope was included as an additional explanatory variable. As observations were not evenly distributed in time, with less observations recorded towards the end of the study, a variable ”hour” was added as an additional continuous fixed effect.Each observed GPS fix was assigned to the hour of its record, with the earliest timestamp registered in each study site being assigned the hour zero. The randomly generated points were randomly assigned to an hour within the determined time continuum of the observed GPS fixes. As our focus was investigating habitat selection at a population-level, we assumed there was no within-dog autocorrelation (space/time) and each dog was independent and exhibited no group behaviour38. Still, to partially account for spatial autocorrelation of each dog’s household, the random effects included in models were defined as each dog’s household geographical location recorded during fieldwork by a GPS device. The restricted maximum likelihood (REML) through Laplace approximations, which can be applied to models with non-Gaussian random effects55, and the Matérn correlation function were used to fit the spatial models with the Matérn family dispersion parameter ν, indicator of strength of decay in the spatial effect, was set at 0.554. More

Study area descriptionYangtze River Basin is situated in central China (Fig. 1). Its geographical coordinates are between 30° 48′ 30″–31° 02′ 30″ N and 112° 48′ 45″–113° 03′ 45″ E. Taizishan is located in the transition zone between the north and south of China, with an altitude of 403–467.4 m. It belongs to the subtropical monsoon humid climate zone and has obvious karst landforms. The farm area is 7576 hectares, the forest coverage rate is 82.0%, and the vegetation is mainly Masson pine, fir, and various broad-leaved tree species. Increased forest coverage reduces sediment production30. The soil is mainly viscous yellow–brown soil and loess parent material. Rain is concentrated in summer, with an average annual rainfall of 1094.6 mm and an average annual temperature of 16.4 °C. Rainfall-related flood risk increased in the Yangtze River Delta in recent years31.The study was based in a Pinus massoniana forest in the Taizishan forest farm of Hubei Province. The Pinus massoniana (Masson pine) is a common species distributed in Central China.Figure 1Geographic location of the study area. Maps were generated using ArcGIS 10.8 for Desktop (http://www.esri.com/software/arcgis/arcgis-for-desktop).Full size imageExperiment designWe chose the Pinus massoniana forest with 47a in the study area as the research object. In the typical Pinus massoniana forest, the separate layers of litter (semi-decomposed and non-decomposed layers) were collected from several 1 m × 1 m quadrat and placed in grid bags. The litter of the semi-decomposed layer have no complete outline, and the color was brown. As the litter leaves of the completely decomposed layer are powdery and are combined with the soil layer, this layer is difficult to collect. Before testing, it was necessary to clean the soil off the pine needles and then allow the litter to dry naturally. The characteristics of the semi-decomposed and non-decomposed litter layers are shown in Table 1. The soil samples need to be dried and screened by 10 mm. When filling the soil trough, every 0.1 m of soil thickness was one layer, for a total of four layers (0.4 m). The characteristics by soil particle sizes are different (Fig. 2). The soil samples were dried naturally, crushed, and then sieved. The soil trough (2 m long, 0.5 m wide and 0.5 m deep) was filled to have a bulk density of 1.53 g·m−3. In this process, an appropriate amount of water was sprinkled on the surface of each soil layer to achieve a soil moisture content consistent with the surrounding, undisturbed, or natural, state. The simulation experiment was conducted in the Jiufeng rainfall laboratory at Beijing Forestry University, China. We used a rainfall simulation system (QYJY-503T, Qingyuan Measurement Technology, Xi’an, China) used a rotary downward spray nozzle. The system is able to simulate a wide range of rainfall intensities (10 to 300 mm h−1) using various water pressure and nozzle sizes controlled by a computer system.Table 1 Characteristics of the non-decomposed and semi-decomposed layers of Pinus massoniana litter.Full size tableFigure 2Soil particle composition of study area soil layers.Full size imageAccording to the results of the field forest investigation, the litter was covered with the experimental treatments shown in Table 2. The treatments mass coverage of non-decomposed litter layer was named as follows: N1 denoted litter mass coverage 0 g·m−2, N2 was ‘the non-decomposed litter mass coverage 100 g·m−2’, N3 was ‘the non-decomposed litter mass coverage 200 g·m−2’, and N4 was ‘the non-decomposed litter mass coverage 400 g·m−2’, N5 was ‘the semi-decomposed litter mass coverage 100 g·m−2’, N6 was ‘the non-decomposed litter mass coverage 100 g·m−2 and the semi-decomposed litter mass coverage 100 g·m−2’, N7 was ‘the non-decomposed litter mass coverage 200 g·m−2 and the semi-decomposed litter mass coverage 100 g·m−2’. N2, N3 and N4 were the undissolved state of litter layer, and N4 (non-decomposed state, ND), N7 (initial stage of litter decomposition, ID), N6 (middle stage of litter decomposition, MD) and N5 (final stage of litter decomposition, FD) respectively represent different stages of litter decomposition.Table 2 The experimental design of this study.Full size tableAccording to the rainfall in the Taizishan area of Hubei Province, erosive rainfall and extreme rainstorms were selected as the research conditions. Summer rainfall events occur mainly in the summer in this area, and a rainfall intensity of 60 mm·h−1 was the most common erosive rainfall intensity. Under extreme weather conditions, the rainfall intensity can reach up to 120 mm·h−1. Our experiments were conducted with 60 and 120 mm·h−1 rain intensities with a rainfall that lasted 1 h. According to the field investigation data of forest land, this area is a low mountain and hilly area with a slope mostly between 5° and 10°. Therefore, 5° and 10° were selected for the slope treatments in this study. The combination of slope and rainfall intensity was named as follows: T1 denoted ‘Slope 5° and rainfall intensity 60 mm·h−1’, T2 was ‘Slope 10° and rainfall intensity 60 mm·h−1’, T3 was ‘Slope 5° and rainfall intensity 120 mm·h−1’, and T4 was ‘Slope 10° and rainfall intensity 120 mm·h−1’. With two rainfall intensities, two slopes, seven litter coverage gradient and two repetitions combined, this study had a total of 56 rainfall events.Experimental procedureBefore the test, the soil samples were wetted for 10 h and then drained for 2 h to eliminate the effect of the initial soil moisture on the soil detachment measurement. When the simulated rainfall started, all the runoff and sediment produced from plot were collected every 5 min in the first 10 min, and then collected once every 10 min during the subsequent 50 min. At the same time, runoff velocity, depth and temperature were measured and vernier calliper (accuracy 0.02 mm) respectively.The overland flow velocity was measured using dying method (KMnO4 solution)32. After judging the flow pattern, we confirmed the correction coefficient K value (in laminar flow state, K = 0.67; transition flow state, K = 0.70; turbulent flow state, K = 0.8). The average velocity of overland flow was obtained by multiplying the correction coefficient K and the instantaneous velocity. Runoff depth was measured using vernier calliper (accuracy 0.02 mm). Runoff temperature was measured using thermometer. When the rainfall experiment finished, the collected runoff samples were measured volumetric cylinder and then settled for at least 12 h. The clear water was decanted, and the samples were put into an oven to dry for 24 h under 105 °C. The sediment sample was dried and weighed with an electronic scale.Calculation of hydrodynamic parametersOverland flow has the characteristics of a thin water layer, large fluctuations of the underlying surface, and unstable flow velocity. At present, most scholars use open-channel flow theory to study overland flow33,34. In open-channel flow theory, the Reynold’s number (Re), Froude constant (Fr), flow index (m), resistance coefficient (f), and soil separation rate (({D}_{r})) are the basic parameters of overland flow dynamics, through Reynold’s number (Re), Froude constant (Fr), flow index (m) can distinguish flow patterns. Re is calculated as:$$Re=Rcdot V/nu ,$$where Re is the Reynolds number of the water flow, which is dimensionless, and can be used to judge the flow state of overland flow. When Re ≤ 500, the flow pattern is laminar; when 500   5000, the flow pattern is turbulent. R is the hydraulic radius (m), which is generally replaced by flow depth as measured by a vernier calliper (accuracy 0.02 mm). (V) is the average velocity (m·s−1); (nu) is the kinematic viscosity coefficient (m2·s−1), and the calculation formula is (nu) = 0.01775·10−4·(1 + 0.0337 t + 0.00021 t2), where t is the test overland flow temperature35.Fr is the Froude constant, which is the ratio of the inertial force to gravity and can be used to distinguish overland flow as rapid flow, slow flow, or critical flow. When Fr  1, the fluid is rapid flow.Fr is calculated as:$$Fr=V/sqrt{gcdot R},$$where (Fr) is the Froude constant of the water flow, which is dimensionless; (V) is the average velocity (m·s−1); g is the acceleration of gravity and has a constant value of 9.8 m·s−2; R is a hydraulic radius (m), and is generally replaced by flow depth as measured by a vernier calliper (accuracy 0.02 mm).Regression fitting is made for runoff depth (h) and single width flow (Q). The runoff depth equation for slope is as follows:$$h=k{q}^{m},$$where q is the single width flow (L·m−1·s−1); h is the depth of water on the slope (m); and m is the flow index, which reflects the turbulent characteristics of the flow state. The larger m is, the more energy the flow consumes in the work of resistance. The comprehensive index (k) reflects the characteristics of the underlying surface and the water viscosity of the slope flow. The larger k is, the stronger the surface material of the slope works on the flow.The resistance of overland flow reflects the inhibition effect of different underlying surface conditions on the velocity of overland flow. The Darcy–Weisbach formula is widely used in research because of its two advantages: applicability and dimensionlessness under laminar and turbulent flow conditions36,37.The resistance coefficient (f) is calculated as follows:$$f=8cdot gcdot Rcdot J/{V}^{2},$$where the resistance coefficient f has no dimension; g is the acceleration of gravity and is always 9.8 m·s−2; R is a hydraulic radius (m), generally replaced by flow depth measured by a vernier calliper (accuracy 0.02 mm); (V) is the average velocity (m·s−1); and J is the hydraulic gradient, which can be converted by the gradient in a uniform flow state and is generally replaced by the sine value of the gradient.Shear stress ((tau)) is the main driving force that affects the stripping of soil particles from the surface soil38. Shear stress is calculated as:$$tau =rcdot gcdot Rcdot J,$$where (tau) is the shear force of runoff (Pa); and r is the density of water and sediment concentration flow (kg·m−3). This study used a muddy water mass and volume ratio in the unseparated state to calculate the density of water and sediment concentration flow.Flow power (W) is the runoff power per unit area of water and refers to the power consumed by the weight of water acting on the riverbed surface to transport runoff and sediment. W is calculated as:$$W=tau cdot V,$$where W is the flow power (N·m−1·s−1); and (tau) is the shear force of runoff (Pa).Soil separation rate (({D}_{r})) refers to the quality of soil in which soil particles are separated from the soil per unit time. The calculation formula is as follows:$${D}_{r}={W}_{d}-{W}_{w}/tcdot A,$$where ({D}_{r}) is the rate of soil separation (kg·m−2·s−1); ({W}_{w}) is the dry weight of soil before the test; ({W}_{d}) is the dry weight of soil after the test, measured by the drying method (kg); t is the scouring time (s); and A is the surface area of the soil sample (m2). More

The attachment site of ticks has been studied in the context of both animal and human tick preference. In Oklahoma, a study of horses indicated that A. americanum preferentially bites the inguinal area, while I. scapularis and D. albipictus, the moose-tick, primarily bite the chest and axillary region, with D. albipictus often being found on the back18. A survey of dogs and cats across the US identified a similar distribution of ticks on dogs, with the attachment being most common on the abdomen, axillary and inguinal regions. However, this was species-specific with D. variabilis preferring the head and neck specifically19. Cats were more successfully parasitized by I. scapularis which preferred the head and A. americanum, which preferred the tail and perianal region19. This is similar to a study of tick distribution on wild black bears (Ursus americanus) in Pennsylvania, indicating that the primary tick present was I. scapularis and that the greatest numbers were found in association with the ears and muzzle20. In these cases, the ability for ticks to attach to specific areas is most likely a result of the grooming habits and abilities of the animals in question.Studies of anatomical region preference in humans also reported tick bite-site specificity associated with particular tick species. For example, in Korea, H. longicornis was determined to prefer abdomen and lower extremities (33%) and the abdomen/inguinal area (26.4%)21, which is a behavior similar to that of A. americanum observed here. Although H. longicornis is present in New York1, insufficient numbers were detected to draw definitive conclusions about its biting preference here. Additionally, a study in England (I. ricinus) reported that tick bites were most common in the legs (50%) of adult humans, but in the head and necks of children (43%)22, a differentiation that our survey does not at this time include. A similar phenomenon was observed in Russia, where tick bites were most common on the head and neck of all individuals (39.2%), but were much more common in children (84.9%)23. This study determined that the bite-site of single tick bites that resulted in infection with the Tick-Borne Encephalitis virus (TBEV) were associated with lethal outcomes if the bites were located on the head, neck, arms or axilla, while less lethality was associated with bites to the lower limbs and groin. This is most directly analogous to the transmission of DTV by I. scapularis, suggesting that bite site may have a similar relationship to disease outcomes in the related North American pathogen/vector pair.Under normal circumstances, ticks exist in sylvatic cycles with specific host preferences based on the tick species and life stage, with spillover to humans occasionally occurring for species with generalist feeding habits. Therefore, the feeding behaviors of ticks are variable, and this influences the ways that the ticks interact with humans.Ixodes scapularis is less specific in host-site preferenceThe primary life stages of I. scapularis that bite humans include nymphs and adult females, although males may also be found on humans. The body segment preference of I. scapularis is less specific than for D. variabilis, which prefers the head, and A. americanum, which prefers the thighs and pelvic region. Ixodes scapularis is primarily found on the central trunk, including the groin/pelvic region, the abdomen, the thoracic region, and the head/neck. This varies between the life stages, with more adults found in the thoracic/abdominal region of the body and nymphs being more commonly found on the arms and legs. This is partly due to the substantial size difference between adult and nymph/larval I. scapularis, with larvae being almost imperceptible and nymphs having a total body length of two to four millimeters. This results in nymphs/larvae being much more difficult to see, allowing them to more readily attach to the most visible portions of the human body while adults are restricted mostly to areas covered by clothing and hair.The presence of ticks on the head and neck indicates that I. scapularis tends to climb, although not with the preference for hair observed with D. variabilis. They appear to spend substantial time moving on the host, a period where they can be removed easily without having had a chance to potentially transmit pathogens by biting. On deer, this corresponds to a preference to move toward the neck and ears where the ticks are more difficult to dislodge24,25. On humans, it results in wide distribution across the whole body with less location specificity than other ticks.In addition to body region and life stage identification, I. scapularis ticks were also screened for several pathogens to determine if infection status influences host site preference. Anaplasma phagocytophilum, B. microti, and other pathogens (DTV and B. miyamotoi) did not influence the body segment the ticks chose to feed. However, in ticks infected with B. burgdorferi, a statistically significant change in the distribution of tick bites marked by an increased report of tick bites in the midsection and a decreased tick bites in the arms, legs, and head. While this may suggest a change in tick behavior/fitness in response to infection, it may also relate to the differences in infection rates of adult and nymph/larval ticks. Larvae, having never fed, are not infected with B. burgdorferi, and the rate of infection in nymphs is lower than that of adults1. Nymphs are less likely to be infected and are more likely to attach to the arms and legs, which is a potential source of the observed difference in infection rates. However, it remains unclear why this is not observed for the other pathogens that follow the same trend of increased infection rate in adult versus nymph/larval ticks.Bacterial and protozoal agents transmitted by I. scapularis take several hours for an infectious dose to be transmitted26,27,28. Therefore, prompt detection and removal of ticks is important for preventing tick-borne disease. Furthermore, understanding where the ticks attach allows them to be more easily detected, and also assists in preparing protective clothing for individuals entering tick-endemic areas. Additionally, knowing the biting location of I. scapularis could aid in detecting potential erythema migrans, a skin condition that occurs at the point of B. burgdorferi infected tick exposure in about 80% of cases29, which is highly diagnostic for both Lyme disease and STARI, which is transmitted by A. americanum.
Amblyomma americanum prefers the thighs and groin of subjectsAmblyomma americanum, the lone star tick, is present throughout the southern portion of New York and is particularly dominant on Long Island1. This species is relatively large, fast, and aggressive, feeding on various animals, including deer, medium-sized animals, and birds30. As a generalist feeder, both adult and nymph/larval A. americanum often bite humans in endemic areas. This experiment identified six larvae, 107 nymphs, and 48 adult A. americanum from human sources. The dominance of nymph submissions is likely due to the large size of the tick, making nymphs and adults easier to spot in more visible areas.In terms of body segment location, all life stages of A. americanum were most often found in the thigh/groin/pelvic region. Considering that most humans encounter ticks while walking through vegetation, the ticks most likely first adhere to the legs and move upward before biting. In this case, the ticks bite rapidly instead of ascending in large numbers to the torso or head. This area is also almost invariably covered in relatively tight-fitting clothing. The closeness of the fabric may also assist in inducing the ticks to feed by slowing their ascent and creating contact to induce biting.While it does not transmit the same range of pathogens as I. scapularis, A. americanum is still a medically significant species. This species can transmit Ehrlichia chaffeensis and E. ewingii31,32, which are at present rare in New York, but are likely to increase as more A. americanum becomes established. Amblyomma americanum is also associated with Southern Tick-Borne Rash Associated Illness (STARI)11, a disease of unknown etiology that has previously been observed in New York33 and with galactose-alpha-1,3-galactose (alpha-gal) allergy, a reaction to the tick’s saliva that can result in a long term, potentially serious allergic sensitivity to the consumption of red meat. While the attachment time required to transmit or induce these pathogens is still unclear, prompt detection and removal of the tick is still recommended. Knowing the approach of the tick and where it is likely to be found improves this process.Additionally, it is unclear if the results observed for A. americanum also apply to the related A. maculatum, the vector of Rickettsia parkeri, a cause of spotted fever. These ticks have been observed in the southernmost portions of New York with a high infection rate with R. parkeri34. Since early R. parkeri infection may result in a visible eschar, understanding where the eschar is most likely located can be critical for rapid diagnosis before the onset of severe disease symptoms. Considering the similarities in behavior between the two Amblyomma species, it may have similar preferences to A. americanum. Other escharotic diseases, such as F. tularensis, may also be present and linked to a tick with a highly dissimilar segment preference. The location of the escar itself, therefore, may be at least partially diagnostic for specific pathogens. However, at present, the sample size within this community engaged passive surveillance program is too small to assess its biting behavior in detail.
Dermacentor variabilis exhibits preference for the human headIn this study, D. variabilis was almost exclusively encountered in its adult life stage. This indicates that while the adult ticks are generalist feeders that may bite humans, the nymph and larval stages are not and have much greater host specificity, either feeding exclusively on a specific type of animal or being restricted to the vicinity of animal burrows. The exact identity of the preferred larval and nymphal host of D. variabilis in New York could not be determined from these data, but is presumed to be one or several rodent species, lagomorph, or mesocarnivore with broad distribution across the eastern United States.Additionally, D. variabilis was unique among the three species of ticks studied here. It had a strong bias toward the head and neck of human hosts, as opposed to a higher preference for the midsection and pelvis/groin with I. scapularis and especially A. americanum. This is clear evidence of climbing behavior, tending upward, but is also indicative of a strong preference for dense hair. In contrast to I. scapularis and A. americanum, D. variabilis in its adult stage is less likely to feed on deer35,36, with a preference for canids36, hence its colloquial name as the “American dog tick”. Hair provides the ticks with the same benefits as feeding on canids. It protects them from being immediately detected and removed, obscuring them until they can feed extensively. This can be of potential medical consequence in the case of tick paralysis, a condition of flaccid paralysis associated with the bite of Dermacentor spp. ticks30. In such cases, prompt removal of the tick is critical for treatment. Therefore, understanding its most likely location can be useful for removal of the tick before the onset of the condition, diagnostically to confirm the presence of the tick, or during treatment to ensure its removal. Considering that the tick will most likely be adult, it should be relatively obvious with careful observation.Limitations of this studyThe data described in this manuscript derived from a set of ticks submitted by general public, with site location from a questionnaire completed upon tick submission. While speciation and pathogen testing were performed under laboratory conditions, the public completed the initial survey and is therefore subject to a level of inherent error and ambiguity. In the context of this study, this mainly concerns whether the body location submitted concerns an attachment or a tick that is still crawling over the potential host in preparation for biting. The term “attachment” may be colloquially interpreted as to contain both categories, or a person can potentially be mistaken about the state of the tick. While ticks filled with blood have fed, the situation is more indeterminate for short-duration attachments where the ticks have not yet begun to engorge. This may introduce some level of error from ticks found on a body segment that were not, at the time of collection, attached. However, the data are overall still useful for predicting the most likely location where ticks of specific species can be found on a person. Studies with test subjects and ticks under controlled conditions may assist in elucidating this matter further. Additionally, this data set was compiled without regard to gender and age group. This data was not collected with this version of the questionnaire; therefore, the tick attachment cannot be stratified by any demographic parameters of tick submitters. More

Study systemOur focal ecosystem is in Selvíria, state of Mato Grosso do Sul, Brazil ((hbox {20}^{circ }) (22′) (41.86”) S, (hbox {51}^{circ }) (24′) (58.90”) W), on a property owned by the São Paulo State University (UNESP). The location covers 350 ha of pasture composed of liverseed grass (Urochloa decumbens). The native vegetation was removed, pasture areas were implemented, and livestock was introduced in the 1970s, maintaining this configuration during the following 50 years. The climate of this area is categorized as equatorial savanna, with dry periods concentrated mostly during the winter, from April to August. During our sampling period (from November 23th, 1989, to November 19th, 2015), no vermifuges and insecticides that could affect negatively the community of dung beetles associated with cow pads were used1.The native dung beetle community at this site is composed of dwellers and tunnelers. Dwellers comprise the Aphodiinae subfamily, whereas all the tunnelers belong to the Scarabaeinae subfamily31. In total, there were eight species classified as dwellers (Ataenius crenulatus, A. picinus and Atanius aequalis-platensis grouped as one species, Blackburneus furcatus, Genieridium bidens, Labarrus pseudolividus, Nialaphodius nigrita and Trichillum externepunctatum) and ten native tunnelers (Ateuchus nr. puncticollis, A. vividus, Canthidium nr. pinotoides, Dichotomius bos, D. semiaeneus, D. sexdentatus, Ontherus appendiculatus, O. dentatus, O. sulcator). These species were chosen for our study because, as the invasive tunneler D. gazella (also from the Scarabaeinae subfamily), they all co-occur in pasture and exploit the same resource (cow pad)32. The initial establishment of D. gazella caused the loss of most of the native tunnelers from the community, with the invader becoming the overwhelming representative of the functional group, and an initial decrease of abundance for dwellers. Differently from native tunnelers, however, dwellers were able to recover their number a few years after invasion (Fig. 1a, Fig. S1).As reported in1, the abundance of dung beetles was significantly affected by both local minimum temperature and relative humidity. The influence of these two factors is expected, as they determine egg and larval survival and development of dung beetles. For example, because dung beetles are poikilotherms, environmental temperature is key to their development and fecundity33. One of the main dweller species, Labarrus pseudolividus, is widely found in locations with temperature averages ranging between (hbox {12},^{circ }hbox {C}) and (hbox {18},^{circ }hbox {C})34, making it tolerant to colder local temperatures. On the other hand, for D. gazella the lower developmental threshold is (hbox {15.5},^{circ }hbox {C}) (individuals cannot survive below this temperature), and the optimum temperature for population growth is (hbox {28},^{circ }hbox {C})35. For both groups, physiological growth and reproduction rates are maintained even when outside temperatures are close to the lower developmental threshold; dwellers, for example, live inside the dung pile, where temperature is higher and less variable than outside36,37. However, while tunnelers oviposit deep in the soil to protect the eggs, warmer and drier conditions reduce dweller egg viability on dung piles since they are exposed38. Low humidity conditions lead to drier dung and can cause egg and insect dessication. In addition, dwellers from our focal system have Palearctic evolutionary origins39; D. gazella’s natural distribution ranges from central to southern Africa40, presenting high physiological plasticity that allows it to tolerate high temperatures and low relative humidity better than other tunneler species41.Functional-group data collection and community structure characterizationDung beetles were collected once a week in a black-light flight intercept trap42, which guarantees the collection of coprophagic beetles. During all collection periods, climate variables were also collected from a meteorological station located within 2 km of our collecting site. See1 for the complete description of the collection process and database. For our purposes, we retained the species, number of individuals per species, and climate variables for each week sampled (Supplementary Information, SI, Figs. S1–S2).We focused first on the weekly abundance data, which we needed to process in order to avoid spurious results in our analyses stemming from the measurement protocol. Specifically, we filtered out seasonal low values associated with sampling in the coldest periods, when few beetles are captured because the reduced activity in all functional groups restricts their spatio-temporal distribution43. Including such samples would not be representative of the community and could bias the analysis since we are investigating community composition (i.e. proportions, very sensitive to low sampling). Thus, we considered only samples with a total number of beetles (that is, summing up all groups together) higher than the value of the median of all data, a conservative threshold that retains observations that allow for as much representation of the community as possible. As will become evident in the Results section and Supplementary Information, less conservative choices for the threshold did not alter our main conclusions.Following Mesquita -Filho et al.1, we categorized all sampled species into either dwellers or tunnelers. D. gazella is a tunneler and, as explained above, the native tunneler species experienced massive declines in abundance after its establishment, leaving D. gazella as almost the single representative in the tunneler functional group during the period of observation1. Thus, given the sharp contrast in community composition, we also separated the data into before and after invasion using to that end the 200th week, when D. gazella was first observed at the study site (September 11th, 1993, starting date for what we will call “after invasion”, our focal period henceforth).To describe community functional composition (i.e. system state) through time, we derived a normalized functional group ratio. First, because the abundance of each functional group spanned up to four orders of magnitude, we performed a logarithmic transformation of the number of captured insects from each group i, (log _{10}(N_{i}+K)), following  Yamamura44. Here, we chose (K=1), but the value of K did not alter our results qualitatively. In addition, the original data showed random mismatches in the phenology of each group, which gave the wrong impression of extreme short-term shifts in functional group dominance within the community. To avoid such artifacts, we used nonparametric local regression (LOESS)45 to smooth the dynamics of each group46. For this smoothing, we employed the loess function in the R software 3.6.147 with a smooth parameter equal to 0.25, but other moderate values (or an optimal value calculated with Bayesian inference by the R function optimal_span) did not alter our conclusions. Finally, we extracted back from the smoothed curve the number of beetles within each functional group to calculate the fraction (f_{dwell}) that measures the relative abundance of dwellers:$$begin{aligned} f_{dwell} = frac{N_D}{N_D+N_T} end{aligned}$$
(1)
where (N_D) corresponds to the number of dwellers per week and (N_T) corresponds to the number of native tunnelers (for the period before invasion), or only the number of D. gazella observed per week (after invasion), using their corresponding smoothed curves. Including also native tunnelers after invasion did not alter our conclusions.Climate driverWe devised a single climatic driver variable that merges the weekly measurement of temperature and relative humidity over the years, abiotic factors key to the survival and reproduction of both groups (see above). We first converted minimum temperatures and relative humidity to normalized climate variables using a min-max normalization (a feature scaling that uses the total range of temperatures or relative humidity, respectively, as normalization factor):$$begin{aligned} T = frac{T_{week} – T_{min}}{T_{max}-T_{min}};;,~ ~ ~ ~ ~ ~ RH = frac{RH_{week} – RH_{min}}{RH_{max}-RH_{min}};;, end{aligned}$$
(2)
where T corresponds to the normalized temperature, (T_{week}) is the weekly temperature, and (T_{max}) and (T_{min}) are the absolute maximum and minimum temperatures observed during the whole sampling period, respectively. We used a similar notation for relative humidity, RH. Based on the information above regarding beetle response to climate, the merged climate factor c was defined as the relationship:$$begin{aligned} c = frac{T}{RH};;, end{aligned}$$
(3)
for (RHne 0). That is, higher temperatures and/or drier conditions (expected to favor D. gazella) lead to higher values for c. On the other hand, lower temperatures and/or more humid conditions (expected to favor dwellers) imply lower values for c. Intermediate values of c can represent either moderate or extreme values for both T and RH.Identifying ecological states and quantifying resilienceWith our (f_{dwell}) data as an index of community composition (i.e. system state), we calculated kernel density functions to interpolate a continuous probability distribution of the relative fraction of dwellers in the community, (p_{n}(f_{dwell})) (function density, R software 3.1.647) for a given range of climatic driver c values. We grouped the (f_{dwell}) data using ranges for c of size 0.4, to ensure a significant amount of weekly samples that allowed for the reconstruction of these probability distributions (see Table S1, first column). Note that bins with extreme values showed few data points (see first and last rows in Table S1), and thus were rejected to prevent misleading results due to reduced sampling. Also note that, for the density function, we used the default Gaussian kernel with a smoothing bandwidth adjusted to be (50%) larger than the default value (“adjust” argument set to 1.5). This conservative choice aims to reduce the effect of the different sampling across c bins and to ensure that differences among distributions across c values are not the result of spurious sampling noise.Further, we transformed the kernel density function:$$begin{aligned} V(f_{dwell}) = -ln (p_{n}(f_{dwell})) end{aligned}$$
(4)
This (V(f_{dwell})) function, called potential (e.g.48), shows by design well-defined minima for the most frequently observed values of (f_{dwell}) (i.e. configurations most frequently observed for the community, which conform the modes of the probability distribution) in a given group of data. At these points, the potential exhibits a change of trend from decreasing to increasing, and therefore its derivative shows a change of sign. Eq. (4), thus, provides a simple criterion to identify possible system states, which is a reason why potentials have been used extensively across disciplines49,50,51. Nonetheless, because the position of extrema is invariant under the transformation, using probability distributions instead would not alter our conclusions.Representing the potential obtained from all the (f_{dwell}) system states associated with a same range of climatic driver c values allowed us to identify stable community configurations associated with a specific climate. The comparison of the potentials obtained for different c ranges enabled the description of how the community changed in response to climatic variation. The location of the minima revealed which states were stable for a given value of the climatic driver; the presence of two minima, then, flagged the existence of bistability (i.e. two different community compositions possible for the same c value).These minima are materialized as wells in the potential’s landscape, which provides an easy way to understand the concept of stability: the dynamics of the system for the given value of the driver will eventually “fall” into a well (either a state dominated by dwellers or a state dominated by tunnelers), with the shape of the well (e.g. its depth) determining how difficult it is for the system to “escape” that state. Therefore, the area inside a well provides quantification of the tendency of a system to stay in that specific state, i.e. the resilience of the associated ecological state or how strong a perturbation has to be to move the system from such an ecological state to another2,3,50,51,52,53. Thus, in addition to number and location of wells, measuring their associated area allowed us to further characterize the resilience of the community. To this end, we first set a visualization window common to all potentials. Specifically, we plotted the potentials within a range for the vertical variable (the potential, V) given by ([-1.5,1.5]); the horizontal variable (fraction of dwellers, (f_{dwell})) is by definition bounded between 0 and 1. For potentials that showed one single well, the area of the well was measured as the area above the potential curve within this visualization window. For potentials that showed two wells (bistability), we measured the value of the potential at the local maximum separating the two wells, and established that value as the upper (horizontal) line closing the area of each well. To ensure all cases were comparable and eliminate any arbitrariness of the choices above, we expressed resilience as a relative area; in other words, we further normalized the well area by the total area across wells for that potential, which means that any single-well case will show a resilience (or relative area) of 1, and the resilience of the two wells when there is bistability adds up to 1.Figure 1Left: Community composition by functional group for all weeks of observation1. Green represents dwellers, blue represents tunnelers, and orange represents the invader D. gazella. Right: Sketch of responses of the community composition to the climatic driver (i.e. phase diagram) expected from the physiological and behavioral characteristics of the functional groups in the community as described in text: linear (red), or non-linear but monotonic without (blue) or with (brown) hysteresis.Full size imageIdentifying ecological transitionsMeasuring a state variable, (f_{dwell}), and a driver, c (order and control parameter, respectively, in the jargon of regime shift theory), allowed us to study how their observed behavior over time materializes in a driver-state relationship (the so-called phase diagram) defining the possible shifts in dominance (i.e. regime shifts) that the community may undergo as climate changes12. The non-monotonic temporal behavior of the components of the order parameter (i.e. dwellers and tunneler availability) and the components of the control parameter (i.e. temperature and relative humidity) makes it difficult to predict the shape of the phase diagram, and therefore whether we can expect alternative stable states in the focal example. For such cases, the dominance of the dung beetle community could (1) shift in a linear fashion toward the functional group favored by climatic conditions; (2) shift between functional groups in non-linear threshold response to climatic conditions without hysteresis; or (3) shift between functional groups in non-linear threshold response to climatic conditions with hysteresis –and thus showing bistability (see Fig. 1b, or12). Other possibilities, e.g. a non-linear shift between functional groups where one group is favored at intermediate climatic conditions12 are discarded as the invader is better suited for warmer and drier conditions. To evaluate which of these possibilities occurred, we represented (f_{dwell}) as a function of c, as well as the location of the minima shown by the potentials above. In addition to the emerging shape of this relationship, this plot can reveal the presence of alternative stable states if two or more different points occur for the same value of the control parameter, c. More

Building from this background the NAM took on these issues as its first-ever grand challenge, as a critical issue of import and urgency for us all. In 2018, the NAM empaneled an international, independent and multidisciplinary commission to create a global roadmap for healthy longevity, complete with evidence-based, targeted and actionable recommendations to move societies forward from an almost-exclusive focus on ‘coping with aging populations’ toward enabling individuals and societies to age successfully, and to reap the economic and societal benefits of longevity. The commission offers a way forward for governments and societies by beginning with recommendations for the next five years, and how these solutions can be financially sustainable through the creation of a virtuous cycle.To support these goals, the commission was to “(1) comprehensively address the challenges and opportunities presented by global aging population; (2) catalyze breakthrough ideas and research that will extend the human healthspan; and (3) generate transformative and scalable innovations world wide”8. The resulting comprehensive report, which was delayed in good measure by the COVID-19 pandemic, was released in June 2022 (ref. 8). We report here a summary of the high-level vision, goals, findings and recommendations of this global roadmap.The evidence for opportunities of longevity and the costs of inactionWe are seeing longer lives with increasing years spent in ill health (that is, the decompression of morbidity)9. The implications of longevity without health are costly ones for the individual, their families and for society. By contrast, scientific evidence shows that the majority of chronic diseases are preventable, and that prevention works at every age and stage of life. Further, the subset of individuals who are the beneficiaries of cumulative health-promoting conditions across the life course are demonstrating healthy longevity, defined as “the state in which years in good health approach the biological lifespan, with physical, cognitive and social functioning, enabling well-being across populations”8. However, only a minority of people in any country have the benefit of the necessary investments that promote health, and disparities in access to these investments across the life course are a major cause of unhealthy longevity. The costs of inaction in the face of widening disparities include the high risk of young people aging with more ill health, and the attendant costs to them and society.Further, the commission reports that when people have health and function in older age, the considerable cognitive and socioemotional capabilities and expertise that accrue with aging, and the prosocial goals of older age, constitute human and social capital assets that are unprecedented in both nature and scale. Contrary to disproven myths, workforce participation not only brings these valuable capabilities (such that intergenerational teams in the workplace are more productive and innovative than single-age-group teams), but older people working is also associated with more jobs for younger individuals10. In the USA and EU, it has been shown that older adults contribute 7% of gross domestic product (GDP) through paid work and the economic value of volunteering and caregiving11, even before opportunities are specifically expanded for the increasing older population. Societies that recognize this potential and invest to create both healthy longevity and the societal organizations and policies through which older adults can contribute to societal good will develop the opportunity for all ages to thrive. The return on investment will be to create older ages with health, function, dignity, meaning, purpose and opportunities — for those who desire it — to work longer, care for others or contribute in ways that they value to their community and future generations.The definition, principles and vision of ‘Vision 2050’ for healthy longevityThe global roadmap builds on the WHO ‘Decade of Healthy Ageing’, the UN Sustainable Development Goals for 2030 and other reports. It sets out principles for achieving healthy longevity using data and meaningful metrics to track achievement of outcomes and guide decision making. The report offers a vision empowered by the evidence: that, by 2050, societies will value the capabilities and assets of older people; all people will have the opportunity to live long lives with health and function; barriers to full participation by older people in society will have been solved; and that older people, with such health, will have the opportunity to engage in meaningful and productive activities. In turn, this societal engagement will create unprecedented social, human and economic capital, contributing to intergenerational well-being and cohesion, and to GDP.Implementing Vision 2050Accomplishing this vision demands ‘all-of-society’ intent — with aligned goals for healthy longevity and transformative action across public, private and academic sectors, and all of civil society and communities — and the implementation of evidence across the full and extending life course. Transforming only one component or sector (for example, health systems) will not be sufficient to create healthy longevity or its full opportunities. Rather, given that nations are complex systems, this vision for our future requires governmental leadership and transformation of all sectors of our complex societal system (Fig. 1).Fig. 1: Relevant actors for an all-of-society approach to healthy longevity.Healthy longevity requires government leadership and cooperation across all sectors. Adapted with permission from figure S-2 of ref. 8.Full size imageInvestment for healthy longevity — across the enabling sectors of health systems, social infrastructure and protections, the physical environment, and work and volunteering contributions — will require intentional planning and leadership to transform those components in tandem, and to resolve disrupters such as ageism, the social determinants of health and inequity, and pollution. These investments across all sectors will create the conditions for achieving healthy longevity and build new capital (human, social and economic) that will benefit all of society. As a result of these investments, society will see younger people thrive and move into a position to age with healthy longevity; those individuals who are already older will be recognized as valuable contributors to society in a ‘pay-it-forward’ stage of life. The underpinning social compact between citizens and government will support valuing each age group’s capabilities and goals, and the building of a society of well-being and cohesion across generations. This is at the center of the virtuous cycle for healthy longevity (Fig. 2)Fig. 2: The virtuous cycle of healthy longevity.Healthy longevity (top) is an outcome of a virtuous cycle, itself contributing to capital development (bottom left). Bottom right, capital (human, financial and social) supports enablers (work, physical environment, health systems and social infrastructure). The enablers propel the cycle, contributing to healthy longevity. Intentional investment for healthy longevity across all enabling sectors will create new capital that will benefit all of society. Adapted with permission from figure 1-4 of ref. 8.Full size imageGoals for initiating the transformation to healthy longevityThe commission identified the following changes that should occur from now to 2027 to start transformation of all of society, towards Vision 2050 and the creation of healthy longevity for all:

Creating social cohesion, social engagement and addressing the social determinants of health through social infrastructure are among the most effective determinants of slowed aging and the prevention of chronic conditions across the life course. Financial security in older age is essential for all.

Governments, the private sector and civil society should partner to design physical environments and infrastructure that are user-centered, and function as cohesion-enabling intergenerational communities for healthy longevity. Initiatives should focus on the inclusion of older people in the design, creating public spaces that promote social cohesion and intergenerational connection as well as mobility, physical activity and access to food, transportation, social services and engagement.

By 2027, governments should develop strategies and plans to arrive at adequately sized, geriatrically knowledgeable public health, clinical and long-term care workforces, and an integration of the pillars of the health system and social services. Together, these dimensions would foster and extend years of good health and support the diverse health needs and well-being of older people.

Governments should work to build the dividend of health longevity in collaboration with the business sector and civil society, to develop policies, incentives, and supportive systems that enable and encourage lifelong learning, and greater opportunities and necessary skills to engage in meaningful work or community volunteering across the lifespan.

We summarize the commission’s recommended goals for each of these sectors in brief in Box 1. Across all sectors, the key first steps that the commission identified are ones that can resolve obstacles to change and plan the change needed to shift multiple complex systems through both top-down and bottom-up approaches, in ways appropriate to each country and context. These initiatives should create enough momentum to foster early returns on investment and optimism to propel sustained investment for subsequent stages. This would need to begin for all governments by 2023, establishing calls to action to develop and implement data-driven, all-of-society plans to build the systems, policies, organizations and infrastructure needed, and for tracking change.Box 1 Goals for 2022–2027 to initiate the transformation to healthy longevityThese goals are reproduced from Global Roadmap for Healthy Longevity8.
Social infrastructure

Develop evidence-based multipronged strategies to reduce ageism against all groups.

Develop plans for ensuring basic financial security for all older people.

Develop strategies to increase financial literacy and mechanisms for promoting working longer, pension options and savings over the life course.

Plan opportunities for purposeful and meaningful engagement by older people at the family, community and societal levels.

Physical environment

At the societal level, improve broadband accessibility to reduce the digital divide and develop public transportation solutions that address first- and last-mile transportation.

At the city level, implement mitigation strategies to reduce the negative effects of the physical environment and related emergencies on older people (for example, air pollution and climate-induced events, including extreme heat and flooding) and design environments for connection and cohesion.

At the neighborhood level, promote and measure innovative policy solutions for healthy longevity, including affordable housing and intergenerational living, zoning and design for connection and cohesion, and the enabling of social capital.

At the home level, update physical infrastructure and policies to address affordability, provide coliving arrangements that match people’s goals and needs, and resolve insufficiencies and inefficiencies in housing stock.

Health systems

Establish healthy longevity as a major goal.

Increase investments in public health systems, which are needed to promote health and prevent disease, disability and injury at the population level, across the full life course. This may require rebalancing investments between this type of public health and medical care, recognizing that such public health is a public good and, as such, tends to be underinvested in.

Provide adequate primary care that includes preventive screening, addresses risk factors for chronic conditions and promotes positive health behaviors, and offers a continuum of medical care, including geriatrically knowledgeable care for older adults.

Make culturally sensitive, person-centered and equitable long-term care systems available, which (to the degree possible) offer dignity and honor people’s preferences about care settings.

Building the healthy longevity dividend

Governments, in collaboration with the business sector and civil society, should design (1) work environments and develop new policies that enable and encourage older adults who want or need to remain in the work force longer, and (2) engagement opportunities that strengthen communities at every stage of life.

Governments, employers and educational institutions should prioritize redesigning education systems to support lifelong learning and training, and invest in the science of learning and training for middle-aged and older adults.

Pilot innovations that incentivize and allow middle-aged and older adults to retool for multiple careers and/or participate as volunteers across their lifespan in roles with meaning and purpose. More