Philippa Kaur

1.Junk, W. J., Bayley, P. B. & Sparks, R. E. The flood pulse concept in river-floodplain systems. In D. P. Dodge [ed.] Proceedings of the International Large River Symposium, Canadian Special Publication of Fisheries and Aquatic Sciences 106, 110-127 https://www.waterboards.ca.gov/waterrights//water_issues/programs/bay_delta/docs/cmnt081712/sldmwa/junketal1989.pdf (1989).2.Karpack, M. N., Morrison, R. R. & McManamay, R. A. Quantitative assessment of floodplain functionality using an index of integrity. Ecological Indicators 111, 106051, https://doi.org/10.1016/j.ecolind.2019.106051 (2020).Article 

Google Scholar 
3.Costanza, R. et al. Changes in the global value of ecosystem services. Global Environmental Change 26, 152–158, https://doi.org/10.1016/j.gloenvcha.2014.04.002 (2014).Article 

Google Scholar 
4.Wohl, E., Lane, S. N. & Wilcox, A. C. The science and practice of river restoration. Water Resources Research 51, 5974–5997, https://doi.org/10.1002/2014WR016874 (2015).ADS 
Article 

Google Scholar 
5.Hamilton, S. K. Wetlands of Large Rivers: Flood plains. Encyclopedia of Inland Waters 607-610 https://doi.org/10.1016/B978-012370626-3.00065-X (2009).6.Opperman, J. J., Luster, R., McKenney, B. A., Roberts, M. & Meadows, A. W. Ecologically functional floodplains: connectivity, flow regime, and scale. Journal of the American Water Resources Association 46, 211–226, https://doi.org/10.1111/j.1752-1688.2010.00426.x (2010).ADS 
Article 

Google Scholar 
7.Waltham, N. J. et al. Lost floodplain wetland environments and efforts to restore connectivity, habitat, and water quality settings on the great barrier reef. Front. Mar. Sci. 6, 71, https://doi.org/10.3389/fmars.2019.00071 (2019).Article 

Google Scholar 
8.Tockner, K. & Stanford, J. A. Review of: riverine flood plains: present state and future trends. Biological Sciences Faculty Publications 29, 166 https://scholarworks.umt.edu/biosci_pubs/166 (2002).9.Erwin, K. L. Wetlands and global climate change: the role of wetland restoration in a changing world. Wetlands Ecology and Management 17, 71, https://doi.org/10.1007/s11273-008-9119-1 (2009).Article 

Google Scholar 
10.Johnson, K. A. et al. A benefit-cost analysis of floodplain land acquisition for US flood damage reduction. Nat Sustain 3, 56–62, https://doi.org/10.1038/s41893-019-0437-5 (2019).Article 

Google Scholar 
11.Quinn, N. et al. The spatial dependence of flood hazard and risk in the United States. Water Resources Research 55, 1890–1911, https://doi.org/10.1029/2018WR024205 (2019).ADS 
Article 

Google Scholar 
12.Pinter, N. One step forward, two steps back on U.S. floodplains. Science 308(5719), 207–208 https://science.sciencemag.org/content/308/5719/207 (2005).13.Kousky, C. & Walls, M. Floodplain conservation as a flood mitigation strategy: examining costs and benefits. Ecological Economics 104, 119–128, https://doi.org/10.1016/j.ecolecon.2014.05.001 (2014).Article 

Google Scholar 
14.Tullos, D. Opinion: how to achieve better flood-risk governance in the United States. Proceedings of the National Academy of Sciences of the United States of America 115(15), 3731–3734, https://doi.org/10.1073/pnas.1722412115 (2018).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
15.Kundzewicz, Z. W., Hegger, D. L. T., Matczak, P. & Driessen, P. P. J. Opinion: flood-risk reduction: structural measures and diverse strategies. Proceedings of the National Academy of Sciences of the United States of America 115(49), 12321–12325, https://doi.org/10.1073/pnas.1818227115 (2018).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
16.Lambin, E. F., Geist, H. J. & Lepers, E. Dynamics of land-use and land-cover change in tropical regions. Annual Review of Environment and Resources 28, 205–241, https://doi.org/10.1146/annurev.energy.28.050302.105459 (2003).Article 

Google Scholar 
17.Entwistle, N. S., Heritage, G. L., Schofield, L. A. & Williamson, R. J. Recent changes to floodplain character and functionality in England. Catena 174, 490–498, https://doi.org/10.1016/j.catena.2018.11.018 (2019).Article 

Google Scholar 
18.Dewan, A. M. & Yamaguchi, Y. Land use and land cover change in Greater Dhaka, Bangladesh: using remote sensing to promote sustainable urbanization. Applied Geography 29, 390–401, https://doi.org/10.1016/j.apgeog.2008.12.005 (2009).Article 

Google Scholar 
19.Amoateng, P., Finlayson, C. M., Howard, J. & Wilson, B. Dwindling rivers and floodplains in Kumasi, Ghana: a socio-spatial analysis of the extent and trend. Applied Geography 90, 82–95, https://doi.org/10.1016/j.apgeog.2017.11.007 (2018).Article 

Google Scholar 
20.Rabalais, N. N., Turner, R. E. & Wiseman, W. J. Jr. Gulf of Mexico hypoxia, a.k.a. “the dead zone. Annual Review of Ecology and Systematics 33, 235–263, https://doi.org/10.1146/annurev.ecolsys.33.010802.150513 (2002).Article 

Google Scholar 
21.Wohl, E. An integrative conceptualization of floodplain storage. Reviews of Geophysics 59, e2020RG000724, https://doi.org/10.1029/2020RG000724 (2021).ADS 
Article 

Google Scholar 
22.Scott, D. T., Gomez-Velez, J. D., Jones, C. N. & Harvey, J. W. Floodplain inundation spectrum across the United States. Nat. Commun. 10, 5194, https://doi.org/10.1038/s41467-019-13184-4 (2019).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
23.Hattermann, F. F. et al. Climatological drivers of changes in flood hazard in Germany. Acta Geophysica 61, 463–477, https://doi.org/10.2478/s11600-012-0070-4 (2013).ADS 
Article 

Google Scholar 
24.Mallakpour, I. & Villarini, G. The changing nature of flooding across the central United States. Nat. Clim. Change 5, 250–254, https://doi.org/10.1038/nclimate2516 (2015).ADS 
Article 

Google Scholar 
25.Corvalán, C., Hales, S., McMichael, A. J., Millennium Ecosystem Assessment (Program), & World Health Organization (Eds.). Ecosystems and human well-being: Health synthesis (World Health Organization, 2005).26.Enhancing Restoration and advancing knowledge of the upper Mississippi river: a strategic plan for the upper Mississippi river restoration program 2015-2025. https://www.umesc.usgs.gov/ltrmp/documents/umrr_strategic_plan_jan2015.pdf (USGS, 2015).27.Nardi, F., Annis, A., Di Baldassarre, G., Vivoni, E. R. & Grimaldi, S. GFPLAIN250m, a global high-resolution dataset of Earth’s floodplains. Scientific Data 6, 180309, https://doi.org/10.1038/sdata.2018.309 (2019).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
28.Sohl, T. L. et al. Modeled historical land use and land cover for the conterminous United States: 1938-1992. U.S. Geological Survey data release https://doi.org/10.5066/F7KK99RR (2018).29.Sohl, T.L. et al. Conterminous United States land cover projections – 1992 to 2100. U.S. Geological Survey data release https://doi.org/10.5066/P95AK9HP (2018).30.Leopold, L. B., & Maddock, T. The hydraulic geometry of stream channels and some physiographic implications. (U.S. Geological Survey, 1953)31.Nardi, F., Vivoni, E. R. & Grimaldi, S. Investigating a floodplain scaling relation using a hydrogeomorphic delineation method. Water Resources Research 42(9), https://doi.org/10.1029/2005WR004155 (2006).32.Di Baldassarre, G. et al. Brief communication: comparing hydrological and hydrogeomorphic paradigms for global flood hazard mapping. Nat. Hazards Earth Syst. Sci. 20, 1415–1419, https://doi.org/10.5194/nhess-20-1415-2020 (2020).ADS 
Article 

Google Scholar 
33.Homer, C. et al. Conterminous United States land cover change patterns 2001–2016 from the 2016 National Land Cover Database. ISPRS Journal of Photogrammetry and Remote Sensing 162, 184–199, https://doi.org/10.1016/j.isprsjprs.2020.02.019 (2020).ADS 
Article 

Google Scholar 
34.Yang, L. et al. A new generation of the United States national land cover database: requirements, research priorities, design, and implementation strategies. ISPRS Journal of Photogrammetry and Remote Sensing 146, 108–123, https://doi.org/10.1016/j.isprsjprs.2018.09.006 (2018).ADS 
Article 

Google Scholar 
35.Jin, S. et al. Overall methodology design for the United States National Land Cover Database 2016 products. Remote Sensing 11, 2971, https://doi.org/10.3390/rs11242971 (2019).ADS 
Article 

Google Scholar 
36. USDA Census of Agriculture Historical Archive http://agcensus.mannlib.cornell.edu/AgCensus/homepage.do;jsessionid=17C0132051BEB31DF79D01B0D07300F2 (US Department of Agriculture, 2007).37.Sleeter, B. M. et al. Land-cover change in the conterminous United States from 1973 to 2000. Global Environmental Change 23(4), 733–748, https://doi.org/10.1016/j.gloenvcha.2013.03.006 (2013).Article 

Google Scholar 
38.Cao, Y. et al. Analysis of errors introduced by geographic coordinate systems on weather numeric prediction modeling. Geosci. Model Dev. 10(9), 3425–3440, https://doi.org/10.5194/gmd-10-3425-2017 (2017).ADS 
Article 

Google Scholar 
39.Piwowar, J. M., Ledrew, E. F. & Dudycha, D. J. Integration of spatial data in vector and raster formats in a geographic information system environment. International Journal of Geographical Information Systems 4, 429–444, https://doi.org/10.1080/02693799008941557 (2007).Article 

Google Scholar 
40.Croissant, C. Landscape patterns and parcel boundaries: an analysis of composition and configuration of land use and land cover in south-central Indiana. Agriculture Ecosystems and Environment 101, 219–232, https://doi.org/10.1016/j.agee.2003.09.006 (2004).Article 

Google Scholar 
41.LaGro Jr., J. A. Land-use Classification (Elsevier Press, 2005).42.Kutcher T. E. et al. Habitat and Land Cover Classification Scheme for the National Estuarine Research Reserve System. (National Estuarine Research Reserve System, 2008).43.Buskey, E. J. et al. in System-wide monitoring program of the national estuarine research reserve System: research and monitoring to address coastal management issues Chapter 21 (Academic Press, 2015).44.Feng, C.-C. & Flewelling, D. M. Assessment of semantic similarity between land use/land cover classification systems. Computers, Environment and Urban Systems 28(3), 229–246, https://doi.org/10.1016/S0198-9715(03)00020-6 (2004).Article 

Google Scholar 
45.Foufoula-Georgiou, E., Takbiri, Z., Czuba, J. A. & Schwenk, J. The change of nature and the nature of change in agricultural landscapes: Hydrologic regime shifts modulate ecological transitions. Water Resources Research 51, 6649–6671, https://doi.org/10.1002/2015WR017637 (2015).ADS 
Article 

Google Scholar 
46.Biondini, M. & Kandus, P. Transition matrix analysis of land-cover change in the accretion area of the Lower Delta of the Paraná River (Argentina) reveals two succession pathways. Wetlands 26, 981–991, https://link.springer.com/article/10.1672/0277-5212(2006)26[981:TMAOLC]2.0.CO;2#citeas (2006).47.Hu, Y., Batunacun, Zhen, L. & Zhuang, D. Assessment of land-use and land-cover change in Guangxi, China. Sci Rep. 9, 2189, https://doi.org/10.1038/s41598-019-38487-w (2019).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
48.Liu, X. et al. High-spatiotemporal-resolution mapping of global urban change from 1985 to 2015. Nature Sustainability 3, 564–570, https://doi.org/10.1038/s41893-020-0521-x (2020).Article 

Google Scholar 
49.Teferi, E., Bewket, W., Uhlenbrook, S. & Wenninger, J. Understanding recent land use and land cover dynamics in the source region of the Upper Blue Nile, Ethiopia: spatially explicit statistical modeling of systematic transitions. Agriculture, ecosystems & environment 165(15), 98–117, https://doi.org/10.1016/j.agee.2012.11.007 (2013).Article 

Google Scholar 
50.Yu, Z., Guo, X., Zeng, Y., Koga, M. & Vejre, H. Variations in land surface temperature and cooling efficiency of green space in rapid urbanization: the case of Fuzhou city, China. Urban forestry & urban greening 29, 113–121, https://doi.org/10.1016/j.ufug.2017.11.008 (2018).Article 

Google Scholar 
51.Yuan, F., Sawaya, K. E., Loeffelholz, B. C. & Bauer, M. E. Land cover classification and change analysis of the twin cities (Minnesota) metropolitan area by multitemporal Landsat remote sensing. Remote Sensing of Environment 98, 317–328, https://doi.org/10.1016/j.rse.2005.08.006 (2005).ADS 
Article 

Google Scholar 
52.Yuh, Y. G. et al. Effects of land cover change on great apes distribution at the Lobéké National Park and its surrounding forest management units, south-east Cameroon. A 13 year time series analysis. Sci. Rep. 9, 1445, https://doi.org/10.1038/s41598-018-36225-2 (2019).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
53.Zhao, J., Yang, Y., Zhao, Q. & Zhao, Z. Effects of ecological restoration projects on changes in land cover: a case study on the Loess Plateau in China. Sci. Rep. 7, 44496, https://doi.org/10.1038/srep44496 (2017).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
54.Rajib, A. et al. Land Use Changes in The Mississippi River Basin Floodplains: 1941 to 2000 (version 1). HydroShare https://doi.org/10.4211/hs.41a3a9a9d8e54cc68f131b9a9c6c8c54 (2021).55.Annis, A., Nardi, F., Morrison, R. R. & Castelli, F. Investigating hydrogeomorphic floodplain mapping performance with varying DTM resolution and stream order. Hydrological Sciences Journal 64(5), 525–538, https://doi.org/10.1080/02626667.2019.1591623 (2019).Article 

Google Scholar 
56.Dottori, F. et al. Development and evaluation of a framework for global flood hazard mapping. Advances in Water Resources 94, 87–102, https://doi.org/10.1016/j.advwatres.2016.05.002 (2016).ADS 
Article 

Google Scholar 
57.Scheel, K., Morrison, R. R., Annis, A. & Nardi, F. Understanding the large-scale influence of levees on floodplain connectivity using a hydrogeomorphic approach. Journal of the American Water Resources Association 55(2), 413–429, https://doi.org/10.1111/1752-1688.12717 (2019).ADS 
Article 

Google Scholar 
58. Climate Change Initiative (CCI) Land Cover products http://maps.elie.ucl.ac.be/CCI/viewer/download.php (2018).59. Land Cover CCI Product User Guide Version 2. Tech. Rep. http://maps.elie.ucl.ac.be/CCI/viewer/download/ESACCI-LC-Ph2-PUGv2_2.0.pdf (European Space Agency, 2017).60.Wilkinson, M. D. et al. The FAIR Guiding Principles for scientific data management and stewardship. Scientific Data 3, 160018, https://doi.org/10.1038/sdata.2016.18 (2016).Article 
PubMed 
PubMed Central 

Google Scholar  More

Influence of soil organic matter on zeta potentialIn this study, the zeta potential of a clayey red soil was compared among 4 types of long-term treatments including manure, NPK + straw, NPK and CK in a subtropical climate. Generally, the manure treatment which also had the greatest concentration of SOC resulted in the highest clay zeta potential (less intense charge imbalance), while NPK + straw did not result in the second highest zeta potential as expected compared to the NPK and CK treatments. Variation in clay zeta potential among types of fertilization might be related with their different SOM content, because SOM had an influence on the zeta potentials via affecting the negative charges of soils19. The zeta potential of manure and NPK + straw treatments having high SOC agreed with earlier studies in Marchuk et al.9 that decreases of SOC via NaOH treatments decreased the negative zeta potential value9, where Claremont soil originally having high SOC (2.2%) displayed a greater degree of decline in negative zeta potential (from − 29 to − 34.9 mV) than Urrbrae having lower SOC (1.4%) (− 66.3 to − 68 mV). However, zeta potential in water dispersible clay responded to SOC contrastly in the study of Melo et al.12 , where Londrina soil with high SOC (5–20 g kg−1) displayed lower negative zeta potential values in water dispersible clay than that in Rondon soil (SOC 5 to 12 g kg−1) in subtropical Brazil.Differences of SOC effect on zeta potential in our study and other studies were probably because ionic strength in bulk solution also affected the intensity of soil charge imbalance. Generally, in tropical and subtropical Ferralsols, high amounts of SOM that was released following the breakdown of macroaggregate provided an excess of negative charges and intensified the imbalance in charge, resulting in more negative in zeta potential of clay12. In contrast to Ferralsols in Brazil, red soil (highly-weathered) in our study showed higher negative zeta potential in manure soils with higher SOM. This was because high ionic strength in bulk solution might counterbalance the negative charges from SOM, and attenuated the imbalance in charges. Hence, manure treatment which provided greater EC and Ca2+, Mg2+ concentration and possibly higher ionic strength was reasonable to allow for more charge balance and greater negative zeta potential values than other treatment.In this study, NPK + straw treatment exhibited similar negative zeta potential values as that in NPK but slightly lower than manure, probably due to the effect of SOM functional group from straw and soil solution concentration. Straw can increase the humin content as reported in the study of Sheng et al.11, and then a decrease of negative zeta potential can be induced as addition of humic acid on a Luvisol20. But the negative humic effect from straw on zeta potential was probably stronger than the positive effect from the increased bulk soil solution concentration in NPK + straw relative to NPK in Fig. 3 where increase of bulk solution concentration was found to increase the negative charge numbers and the negative zeta potential in Ultisol and Oxisol15. Therefore, our hypothesis that organic treatments decreased negative zeta potential value of soil was not supported for manure treatment, but was for NPK + straw treatment.NPK + straw’s similar effect on negative zeta potential as NPK treatment was probably also related with their similar pH values. The effect of pH on the potential of clay surfaces can be related to the amount of variable charge on the external surface of the clay particles. Negative zeta potential decreased with rising pH of the solution due to deprotonation of the functional groups on the surface of the organic matter and Fe/Aloxides in NPK + straw treated soils. An increase of soil pH (from 3.5 to 7.5) influenced zeta potential through production of more negative net surface charges on soils in subtropical Australia21,22. Therefore, the pH in our study after KCl adjustment that showed a first increase and then decrease pattern with the increase of concentration, can help to explain the bell shape pattern of negative zeta potential (first decrease and then increase). However, in our study, the pH pattern with increment of KCl concentration was different from the results in study of Yu et al.8 where a continuous decline pattern in pH of two soils (Vertisol and Ultisol) was reported when the KCl concentration increased from 10–5 to 10–1 mol L−1. This is probably because the Ultisol possessed high amount of variable charges from Fe or Al oxides, which resulted in the diffusion layer attracted more positive charged cations (i.e. K+) from bulk solution to balance the increased negative charge on the surface of colloidal particles in order to maintain the electrical neutrality of the system15. This indicated that when KCl concentration was low, between 0 and 10–2 mol L−1, part of K+ was attracted to the diffuse double layer and the remaining K+ hydration allowed for raising in soil pH. When KCl concentration was beyond 10–2 mol L−1, many Al3+ions on soil exchange site were released into solution (0.03 to 0.12 mg L−1) through K+ exchange and probably dropped soil pH (data not shown).Studies also found that the effect of SOM on zeta potential of clay also varied for soils in different climate. Yu et al.8 compared rice straw incorporation effect on two soils (Ultisol and Vertisol) and found that similar SOC content resulted in contrasting effects on surface potential of two types of soils, where surface potential of Ultisol continuously increased while firstly increased and became stable for Vertisol with increase of treated solution concentration. Different SOM effect on soil potential properties of two soils were probably associated with presence of soil variable charges in Ultisol23. SOM and Fe/Al (hydro)oxides in Ultisol carried a larger number of variable surface charges, and resulted in a strong overlapping of oppositely charged electric double layers (EDLs) between SOM and Fe/Al (hydro)oxides at low concentration8. The overlapping of oppositely charged EDLs between SOM and Fe/Al probably yielded in an increase in negative surface charge for Ultisols compared to Vertisol.Effect of SOM and zeta potential on soil aggregationIncrement in content of SOM after additions of straw or other organic treatments can improve aggregate stability6,24,25. The hydrophobic organic compounds that coated around soil particle can act as nucleus of aggregate formation and reduce the destruction effect from water infiltration26,27. The hydrophobic-C/hydrophilic-C increased from 1.04 to 1.07, from 1.22 to 1.27 for chicken manure and maize residues treatments, respectively, when soil water conditions changed from water deficiency to natural rainfall treatment28. This indicated that a small change of hydrophobic-C/hydrophilic-C might result in substantial change in soil water, which was a critical factor of aggregate development28. Xue et al.24 also reported that a small difference of aromatic percentage between tillage + straw and no tillage + straw treatments resulted in significant differences for aggregate ( > 0.25 mm). Hence, small variation in soil hydrophobic-C groups can yield in soil aggregate variation. In our study, the manure treatment, which had higher SOM and hydrophobic-C (aromatic C) while lower hydrophilic-C than other treatments, was probably reasonable to yield in its higher stability than others. In these previous studies, the positive effect of SOM on soil aggregate development was attributed to the increment in van der Waals force between soil particles. However, different from our study, Melo et al.12 reported that Londrina soil with high SOC released greater water dispersible clay (60–80%) than that in Rondon with low SOC (50–70%) after mechanical breakdown of macroaggregate. This was probably due to the repulsive force prevailing attractive force between soil particles as affected by more negative zeta potential or surface potential8.Clay zeta potential influenced the powerful electrostatic fields, soil internal forces and aggregate stability9. Decrease in negative clay zeta potential mainly yielded an increase in the soil microaggregate portion ( More

Based on comparative analysis of aerial imagery, the marshes in the Florida Panhandle were overwhelmingly undamaged by Hurricane Michael. Of the 173,259 km2 of marsh analyzed across Bay, Gulf, and Franklin counties, only 1.9% (3371 km2) was classified as damaged after Hurricane Michael. Less than 2% of marshes were damaged in Bay and Franklin counties, and less than 5% of marshes were damaged in Gulf County. This result suggests that the majority of marshes in the study counties can withstand the effects of a category 5 hurricane and continue to provide coastal protection ecosystem services29. This also identifies marshes as more resistant than most other coastal defenses, including bulkheads. While this study did not examine bulkhead failure, similar studies found that 76% of bulkheads surveyed in North Carolina were damaged after Hurricane Irene29, and a large portion of bulkheads in the Florida Keys experienced significant damage after Hurricane Irma31,32.Damage to marshes in the Florida Panhandle from Hurricane Michael was spatially distributed largely due to the storm track. Shortly before making landfall in Mexico Beach, the storm turned northeastward, influenced by the southern edge of mid-latitude westerlies1. The highest hurricane wind speeds are typically associated with the front-right quadrant of the storm and weaken rapidly after landfall, which was also true for Hurricane Michael20. As expected, the majority of damaged marshes (4.3%) were in Gulf County, which is directly southeast of landfall. Given the sizable decrease in wind speed and storm surge as the storm lost intensity over land, about half of the damage (51.6% across all counties) occurred within 50 m of the coast, and the vast majority of damage (95.1% across all counties) occurred within 500 m of the coast.The types of damage were consistent with hurricane impacts in the Gulf of Mexico, with sediment deposition and wrack deposits among the most common morphological impacts from hurricane strikes21,22. The type of damage was also spatially distributed. Marshes in Bay County experienced the highest wind speeds and also experienced t he most damage from fallen trees, a largely wind-driven form of damage. Marshes in Franklin County experienced the largest inundations, as well as the most conversion to open water. Deposition of sediment or vegetation can be influenced by both wind and inundation, whether through aeolian transport3,4,5,7,8,23 or overwash12. Therefore, it is not surprising that deposition was both the most common as well as the most evenly spatially-distributed form of marsh damage. Vegetation loss, though less common than deposition, is also largely driven by high-velocity wind-driven currents; a particularly intense flow can pluck or denude a marsh of its vegetation24,25.Marsh damage is also likely closely tied to physical properties of the marsh, in addition to the spatial component of storm conditions. The vast majority (88.5%) of marsh damage occurred at elevations of less than 1 m (NAVD88), consistent with both average marsh elevation and areas of greatest susceptibility to inundation. Localized susceptibility to damage may also indicate differences in marsh soil composition or shear strength24,26 or previous physical disturbance27. Given this study’s reliance on visual observations to determine damage, it is important to consider that not all storm-induced marsh damage can be ascertained visually; subsurface processes such as shallow subsidence or expansion can greatly influence the marsh’s elevation and resilience14,28.It is also important to consider that, of the ~ 2% of marshes in the Florida Panhandle that were damaged, ~ 80% experienced deposition of sediment or vegetation on the marsh surface, which is a potentially less-permanent form of damage, especially compared to fallen trees or conversion to open water. Hurricanes regularly place thick sediment deposits on marshes in the Gulf of Mexico; short-term sedimentation rates in coastal Louisiana marshes after Hurricane Andrew increased for three months post-storm by 1–3 orders of magnitude compared to pre-storm rates, suggesting that resuspension and deposition continued well beyond the timeframe of the passing storm33. Deposition of sediment from storm overwash may actually benefit the system, increasing total marsh elevation and counteracting sea-level rise8. While storm-induced sedimentation may increase resilience, deposits that are too thick ( > 5–10 cm34) may cause plant mortality and ultimately reduce the stability of the marsh and its resilience to storm impacts.Much of the deposition on the marsh surface was rafted vegetation and wrack mats, which impacts marsh vegetation differently than sediment. This wrack deposition is consistent with past hurricane impacts in the Gulf of Mexico; wrack deposits from Hurricane Andrew completely buried vegetation in coastal Louisiana marshes, and areas of especially thick wrack were slow to recolonize, even a year after landfall35. A 1995 study, however, found that only 30% of wrack mats on a New England salt marsh damaged underlying vegetation36, and while wrack mats were a major cause of damaged low-marsh vegetation in a Virginia salt marsh, vegetation typically recovers37. This suggests that even the 2% of damaged marshes in the study area may be more resilient than originally thought.Though the vast majority of marshes in the Florida Panhandle were not visually damaged by Hurricane Michael, of the damaged marshes with aerial imagery from six months after the storm, only 16% recovered. Marshes exposed to less extreme environmental conditions during the storm often were more likely to recover; this was the case for the relationship between mean maximum wind speed and deposition recovery, as well as maximum inundation and vegetation loss and conversion to open water.Perhaps unsurprisingly, some forms of damage are easier to recover from than others. More than 45% of marshes experiencing vegetation loss had recovered within six months, whereas only 16% of marshes experiencing deposition of sediment or wrack, the most common form of damage, had visually recovered within 6 months. This is consistent with other studies; marshes in the Mississippi River Delta denuded by Hurricane Camile partially or completely recovered within one growing season, provided the root mat was not destroyed and the area was not permanently submerged12. This study is limited to 6 months after landfall, which is a relatively short time frame and does not include the spring and summer growing seasons. Given that other studies have shown that it can take over a year to recover marsh vegetation after disturbance (including storms and fire)38, and it is likely that more marsh recovery would be observed a full year after disturbance. Despite these considerations, this study provides important insight into the short-term recovery of vegetation (through recolonizing after plucking or growing through thin layers of deposition).Once vegetation is substantially disturbed or submerged, however, recovery becomes more difficult, and it is these areas that might be prioritized for more active recovery interventions. Kirwan et al. (2008) found that disturbed areas experience decreased vertical accretion, which in turn causes localized submergence of the marsh platform, as well as potential expansion of channel networks, further destabilizing the marsh. Less than 4% of marsh that had been converted to open water and none of the marsh experiencing channel widening or cutting had recovered within 6 months. This is consistent with damage identified from a survey of hurricanes impacting southern Louisiana over the last 50 years12. Since newly formed or widening ponds or channels indicate significant loss of the marsh platform, it is far more difficult for the marsh to both recover elevation and regrow vegetation. Research has shown that ponds created by hurricane impacts could maintain their shape for decades and eventually become a permanent feature39.This study found that though the vast majority of marshes in Bay, Gulf, and Franklin counties were not visually damaged during Hurricane Michael, of the 2% of marshes that did experience damage, 84% had not recovered within six months. This suggests that marshes may be largely resistant to storm impacts but not particularly resilient19, at least within the first six months of disturbance. Recovery rates for marshes on public land were significantly higher than private land. This speaks to the benefits of public land management and the need to create better incentives for marsh management on private lands. Additionally, the majority of marshes on public land are within larger wildlife refuges, state parks, or management areas and, as a result, may be less altered and more resilient to storm impacts.Future damage surveys incorporating on-the-ground assessments and finer-resolution aerial imagery are recommended to fully understand the impacts of intense storms on salt marshes. Since this study, by design, only quantifies damage and recovery visible from aerial imagery, it is important to consider that there may be additional damage to the marshes that may be small, subtle, or belowground and therefore not observable at the scale of the imagery. While this study is specific to the inundation and hydrodynamic conditions associated with Hurricane Michael, the implications about marsh performance and recovery provide an important base for similar future assessments of future storm impacts.Post-storm management actions may be key to increasing marsh resistance and resilience after storm impacts. As seen in this study, deposition is the one of the most common forms of marsh damage, but also one of the most easily reversible. While storm-induced sediment deposition is often largely beneficial to building marsh elevation31, wrack deposits may damage vegetation36,37, which in turn must recover by the next growing season to remain resilient to the next hurr icane season. Removing the thickest layers of wrack buildup to prevent vegetation death is a relatively straightforward management activity, which in turn will build the marsh’s resilience to future near-term storm impacts. Cost-effective restoration and recovery efforts should prioritize maintaining vegetation health, particularly in locations where the marsh platform is still intact post-storm. Post-storm wrack removal is a target for recovery funding that may be particularly beneficial.Results of this study inform policy, funding, and approaches for hazard mitigation, disaster recovery, adaption, and conservation. Very few marshes were damaged or failed during a direct hit from one of the strongest hurricanes to impact the continental United States, and marsh failure rate was generally much lower than that seen for other artificial defenses, even in lower-intensity storms30,31,32. A portion of damaged marshes also began to recover within six months with few or no interventions, which artificial defenses cannot. Managed marshes on public lands are more likely to be both resistant and resilient to storm damage. Previous work has shown that marshes significantly reduced hurricane damage to property15 and are a cost-effective method of coastal protection40. Combined with our results, which indicate that marshes are highly resistant to storm impacts, these studies identify the need for additional public and private incentives, including insurance incentives and hazard mitigation funding, to conserve and restore marshes for their benefits to both people and property. After a storm, allocation of recovery funds to this natural, national infrastructure is imperative to improving management and recovery trajectories, particularly given the potential for more intense hurricane landfalls as the climate changes19. More

Study siteThe study covered the Brazilian Atlantic Forest domain (Fig. 2), which is located on the east coast of Brazil between latitudes 5° and 30° south, expanding over 500 km inland in the south. It consists of a total area of 1,085,151 km2 with limits defined by the Brazilian Ministry of the Environment48. The total area covered by Sentinel-2 tiles overlapping with the Atlantic Forest domain is ~2,006,959 km2. This latter area was used to compute the descriptive statistics of detections.DataSentinel 2 imagesThe pink or magenta blossoms of Pleroma trees were mapped using Sentinel-2 multi-spectral data with 10 m spatial resolution taken approximately every five days under the same viewing conditions. We used only Sentinel-2 images with Level-1C correction—which are orthoimage products, i.e. map projections of acquired images using a digital elevation model to correct ground geometric distortions—and delivered in images of 100 km × 100 km. Pixel radiometric measurements were provided in Top-Of-Atmosphere (TOA) reflectances (coded in 12 bits)49.In the analysis, 213 Sentinel-2 tiles covering the Brazilian Atlantic Forest domain were used, totaling 2,006,959 km2 which is equivalent to ~20 billion Sentinel-2 pixels with 10 m spatial resolution (Fig. 2a). Amongst the 213 selected tiles, 36 had 2 orbits to download to obtain the full tile image due to the overlapping orbit paths (called replicates in the following text).For each tile and replicate (213 + 36), the times series between 31 June 2016 and the 1 July 2020 was downloaded from the Google Cloud Storage Sentinel-2 repository (https://cloud.google.com/storage/docs/public-datasets/sentinel-2). To reduce the dataset size, we retained only images with less than 80% cloud cover; and, from the month outside the flowering months of the Pleroma trees (July to November), we kept only images with less than 25% cloud cover. The complete dataset was made up of 33,798 Sentinel-2 images.Four spectral bands available at 10 m spatial resolution were used: Red (665 nm), Green (560 nm), Blue (490 nm) and NIR (842 nm). A border of 120 pixels with NA values was added to the image to produce images of 10240 × 10240 pixels to ease automation of the image analysis workflow, which generally works with 2n × 2n size pixel images. In our case here, the deep learning analysis was made with 128 × 128 pixel images and an additional 8 × 8 border. Sentinel L1C reflectance values are in the range of 0–10000 and were converted to 8 bits (0–254) with the following rules : for Red, Green and Blue bands, we kept the minimum value between 2540 and the original pixel value, divided this value by 10 and converted the result to integer; and for the NIR band, we keep the minimum value between 2540 and the original pixel value divided by a constant equaling 3.937, divided this value by 10 and converted the result to integer. While it was not expected to have RGB pixel values for vegetation with reflectance above 2540, it occured frequently for the NIR values. Dividing the NIR band values by the constant 3.937 enabled scaling the full range of the original NIR values between 0 and 2540 without losing too much information. For each tile, all 4 bands were saved in one GeoTIFF of 8 bits to ease storage and processing. The size of the complete dataset was 5.59 teraoctets. The automatic download, scaling and conversion of the images to 8 bits took about 25 days (from 16 July 2020 to 3 August 2020 and from 10 September 2020 to 13 September 2020).Environmental dataFigure 12Environmental and climatic variables used in the study to analyse spatial distribution of Pleroma trees (a) elevation (m), (b) slope (°), (c) tree cover (%), (d) mean annual precipitation (mm yr−1), (e) annual mean of minimum temperatures (°C), and (f) maximal annual temperatures (°C).Full size image
To test the association of Pleroma trees with elevation and slope, elevation data from the Shuttle Radar Topography Mission (SRTM) were used50 (Fig. 12a). Specifically, we used the 3 arc-seconds (~90 m) spatial resolution digital elevation database (version 4) provided by the CGIAR Consortium for Spatial Information51. This dataset, in comparison to the original NASA STRM dataset, has been processed to fill data voids. From this dataset, we used the variables elevation (m) and computed slope (°) considering the four neighbor pixels (Fig. 12b). To analyse the relationship between Pleroma trees presence and forest tree cover, we used the tree cover percentage for the year 2000 at 30 m of spatial resolution, which we obtained from the global forest cover dataset (Fig. 12c), which is based on Landsat time series52.The association of Pleroma trees with local climate was tested using the annual means of precipitation and air temperatures (Fig. 12d–f). The mean annual precipitation over the study period was computed from the CHIRPS v2p0 monthly precipitation dataset at 0.05° of spatial resolution produced by University of California, Santa Barbara (UCSB). CHIRPS data are global rainfall estimates from rain gauges and satellite observations53. The mean of maximum and minimum air surface temperatures over the study period were computed from the Aqua/AIRS L3 Daily Standard Physical Retrieval (AIRS-only) at 1° of spatial resolution V7.0 (AIRS3STD). AIRS, the Atmospheric Infrared Sounder on NASA’s Aqua satellite, gathers daily infrared energy emitted from Earth’s surface and atmosphere globally and provides 3D measurements of temperature and water vapor through the atmospheric column54. The annual mean of minimum and maximum air surface temperatures was calculated using the daily air surface temperature measured from the descending orbital pass, which occurs at 1:30 am local time (’SurfAirTemp_D’), and the ascending orbital pass, which occurs at 1:30 pm (’SurfAirTemp_A’).Additionally, maps produced by the Brazilian Institute of Geography and Statistics (IBGE) of the geomorphological units and rivers of Brazil were used to describe the spatial distribution of the blossom detections33.All environmental variables were resampled to a raster of 1280 × 1280 m spatial resolution using an average interpolation to match the resolution of the Pleroma tree detection dataset.ModelNeural network architectureThis detection model is a deep learning neural network (Fig. 13), more specifically an encoder with a VGG16-like structure35, that given an image (input image) return the probability of presence of Pleroma trees with flowers in the image. The model inputs are 4 bands RGB-NIR images made up of 136 × 136 pixels at 10 m of spatial resolution (Fig. 13). Sentinel-2 tiles of 10240 × 10240 pixels were cropped based on a regular grid of 128 × 128 pixels, and 4 neighbouring pixels were added on each side to create an overlap between the patches. The resulting images are 136 × 136 pixels in size. However, in the training, the presence or absence of blooming Pleroma was given only for the images of 128 × 128 pixels without consideration of the borders. This enable to avoidance of the border effect that is common in convolutional neural networks. Each image of 136 × 136 pixels goes through a data augmentation process that consists in random vertical and horizontal flips. No additional data augmentation necessary due to the natural data augmentation provided by atmospheric conditions and illumination. After data augmentation, the images were then fed to the detection encoder. The encoder was made up of 5 consecutive convolution and pooling blocks, one fully connected layer (dense 100) and a final output layer with a softmax activation that provided the probability of presence of blooming Pleroma trees in the image (Fig. 13). Additionally, one drop-out layer was used at the end of the fully connected layer to perform further implicit data augmentation and avoid overfitting during training. The model has a total of 25,448,686 parameters, of which 25,440,622 are trainable. The model was coded in R language55 with Rstudio interface to Keras and TensorFlow 2.256,57,58,59.Figure 13Architecture of the Pleroma blossom detection model.Full size imageNetwork trainingTo make the training sample, a manual sample was produced for the Sentinel-2 tile 23KMQ, in the area where we had previously made a high resolution map of blooming Pleroma6, and for five other tiles where flowering Pleroma were detected visually from high resolution Google Earth images (22JFQ, 22JGQ, 23KLP, 23KLQ and 23KNQ, respectively). What is identified in the Sentinel-2 images are forest stands dominated by Pleroma and not single individuals. Pleroma trees have a small stature (8–12 m height) and crown of less than 10 m and one tree alone cannot influence sufficiently the reflectance to be clearly detectable in Sentinel-2 images. However, they occur very frequently clumped together, a common behaviour of this pioneer Genus. These flowering Pleroma dominated forest stands were easy to identify visually in the Sentinel-2 images because they combined several very distinctive features. First, an intense magenta-to-deep-purple color, which is an unusual color for other land covers in this ecological domain. Second, these identified Pleroma pixels should be undoubtedly identified as forested pixels and have a green color outside the blooming season. Third, Pleroma dominated forests often formed continuous magenta-to-deep-purple patches of size ranging from some 10 m × 10 m pixels to more than thousands of pixels and the shape of the patches tend to present linear features, likely representing the border of the space that was colonized by the Pleroma trees. Fourth, individual crowns were not visible, and the texture of the patches was very smooth during the blooming season with sometimes some inclusions of tree crowns of green color. Finally, texture of the Pleroma dominated forest stands outside the blooming season shown a smooth green texture, more homogeneous and with less shade than other forests. The first sample was constituted of images of background and of blooming Pleroma dominated forest stands that were following the previously described criteria. From this sample, we train a first model and applied it to the complete time series of Sentinel-2. From the results of this model, we obtained a first map of Pleroma trees and were also able to identify the main detection errors of this model, mainly clouds and dirt roads proximity with some unidentified agriculture fields or sometimes Eucalyptus plantation. The results of this first model were checked visually to produce a second larger sample (which was used for the results presented in this study) made up of images containing blooming Pleroma dominant to monospecific forest stands, a set of background images without blooming Pleroma and images identified erroneously by the first model as containing blooming Pleroma. While a large majority of the detected pixels were undoubtedly forest stands dominated by Pleroma trees, some other isolated trees of the genus Handroanthus (Ipê in Brazil or Lapacho in Argentina) with pink flowers and large crowns covering several pixels of 10 m × 10 m were also detected and kept in the training sample. For these particular Handroanthus trees, crowns were visible during and sometimes also outside the blooming season, which was not the case for detected Pleroma dominated forest stands. Finally, as our model detected also large Handroanthus trees, we must acknowledge that other tree species with highly similar features could also potentially being detected.The final training samples comprised a total of 158,612 images of 136 × 136 pixels. Among them, 35,541 contained blooming Pleroma trees and 123,071 images contained only background. Among the background images, there was nine different images types: images without blooming Pleroma, i.e., background such as other land covers, urban structures, water surfaces and agriculture and other land uses (57,007), images with forests containing Pleroma but outside the flowering period (23,427), images with clearly identified detection errors mainly located in the east of the São Paulo state (12,965), clouds and detection errors in clouds (10,991), images clearly identified as detection errors near the State of Bahia (9030), other detection errors over Atlantic Forest (5843), images of forests without Pleroma trees during the season of blooming (2170), images with identified detection errors in the northern part of Atlantic Forest (1126) and images with no data (512). Of these images, 80% (126,890) were used for training and 20% (31,722) used for validation.During network training, we used a standard stochastic gradient descent optimization, a binary cross-entropy loss and the optimizer RMSprop60 with a learning rate of 1e-4. We used the accuracy (i.e. the frequency with which the prediction matches the observed value) as the metrics of the model. However, due to the imbalance between the number of blooming Pleroma and background images, the metric of the model was weighted by one for the background and, for the Pleroma, by the ratio between the number of background images and the number of images containing blooming Pleroma: that is, ~3.5. The network was trained for 5000 epochs, where each epoch was made of 61 batches with 2048 images per batch and the model with the best weighted accuracy was kept for prediction (epoch 4331 and weighted accuracy of 99.58%). The training of the models took approximately 9 hours using a Nvidia RTX2080 Graphics Processing Unit (GPU) with an 8 GB memory.PredictionTo avoid border effects, each 10240 × 10240 pixels Sentinel-2 image was cropped on a regular grid of 128 × 128 pixels (1280 × 1280 m), and 4 neighboring pixels were added on each side to create an overlap between the patches. The function gdal_retile61 was used for this operation. Prediction was then made for each subset image: for each image, the detection model returned 0 or 1 if a blooming Pleroma was found in the image. Then the results were spatialized again using the grid, but this time, each cell of the grid only received 1 value, the prediction, resulting in a raster of 80 columns and 80 rows and a spatial resolution of 1280 m, of the same extent as the Sentinel-2 image. The value of the pixels (1 or 0) indicated the presence or absence of blooming Pleroma trees in this squared area of 1280 m of side. Prediction using GPU of a single tile of 10240 × 10240 pixels took approximately 1 minute on a Nvidia GTX1080 with an 8 GB memory and 45 s on a Nvidia RTX2080 with an 8 GB memory. The prediction for the complete Sentinel-2 time series presented in this work took approximately 22 days using a Nvidia GTX1080 GPU—from the 30 October 2020 to the 20 November 2020.Spatio-temporal analysisTo analyse the seasonality of the detections, daily maps of flowering Pleroma detections were produced for the studied period on a grid overlapping the entire Atlantic Forest (projection UTM 23S and datum WGS84) with a spatial resolution of 1280 m to match the resolution of the predictions. For each day, each pixel of the grid was given a classification: observed with flowering Pleroma, observed without flowering Pleroma, observed with clouds (using the cloud cover mask for Sentinel-2 images of this day) or as non-observed (no image or NA data for the pixel on that day). These daily grids were use to produce the map of flowering Pleroma trees (number of detections of flowering Pleroma for each pixel along the time series), the map of the total number of observations per pixel and the map of the total number of observations without clouds.To analyse the seasonality of blooming, the detection results were aggregated by month—even with a 5-day frequency there were still too few observations to analyse each annual timing and duration of flowering, and changes of the flowering dates between years were not expected based on the existing botanical information of the species. For each pixel, the number of detections per month were divided by the total number of observations without clouds per month. This enabled to normalized the detection values between zero and one and made sense given that we were not interest in the number of detections but rather in the times of the year when the number of detections was the highest: the peak of the blooming.To find the characteristics of these time series—one or more blooming peaks and the days when the blooming begins, peaks and stops—the normalized time series of mean monthly observations of flowering Pleroma were filtered using the Fourier transform (FT) (Eq. 1). This decomposition was made the keeping only the annual, bis- and tris-annual frequencies that compose the blooming signal, and to provide a continuous representation of the discrete blooming observations. In other worlds, the Fourier transform of the normalized time series observations enabled to model and compute the values of blooming for each day of the year and better estimate the days blooming started, peaked, and ended. While initially, a decomposition with only annual and biannual frequencies was expected to fit well to the times series (as more than two peaks per year were not expected), it appeared that when the two peaks were close in time (such as in a 2–3 month interval), only annual and biannual frequencies were not sufficient to give a good model of the signal, and the triannual frequency was added to resolve this issue. Furthermore, it was assumed that other periods in the signal were only constituted by noise.The blooming signal was modelled by the following equation:$$begin{aligned} {widehat{bloom}}(t)& = bloom_0 + pow_0 ,left( p_{4} sin left( 2pi frac{1}{4} t + rho _4right) right. \ & quadleft. + p_{6} sin left( 2pi frac{1}{6} t + rho _6right) + p_{12} sin left( 2pi frac{1}{12} t + rho _{12}right) right) end{aligned}$$
(1)
with (p_4 + p_6 + p_{12}=1) and for (t=1,ldots ,12 times n), ({widehat{bloom}}) is the filtered blooming time series; (bloom_0) is as an estimate of the mean annual blooming; t is the time in months; (rho _4), (rho _6) and (rho _{12}) are the delay of signal components with periods of 4 months, 6 months and 12 months, respectively; (pow_0) is the power of the signal and (p_4), (p_6) and (p_{12}) are the relative proportions in the signal of the periods of 4 months, 6 months and 12 months, respectively.To ease optimisation and cohere with the biological significance of our model, some data cleaning and adjustments were made. First, pixels with less than 4 observations over the 4-year period were removed from the analysis. Second, isolated peaks with only 1 or 2 observations during the 4-year period and between months without Pleroma detection were set to 0. Third, all the values of the normalized blooming time series were multiplied by 10, which seemed to ease the convergence of the optimisation algorithm. Fourth, the first months before and after the blooming period were set to a negative value equal to − 0.15 × the maximum value of the pixel time series. This was made based on the assumption that blooming is quite fast (based on the observation data) and happens between the month when the blooming is first observed and the previous month (and when blooming is last observed and the next month), and it forced the model to go below the 0 value during this period. Fifth, a weight was added to each point corresponding to its value, as we were interested in estimating accurately the peak value. A weight value of 0 was set to the month with a 0 value, and a weight of 1 was set to the months with negative blooming values (pre- and post-peak months). Finally, to facilitate the optimization, the time series of values and weights was replicated 3 times (n = 3). While this did not change the periodicity of the signal, it enabled to estimates better the value of the first and last month of the time series, as well as to ease optimisation. The parameters (bloom_0), (pow_0), (p_4), (p_6), (p_{12}), (rho _4), (rho _6) and (rho _{12}) were then estimated by a weighted least square minimization using the weights previously described. The accuracy of the model was given by the weighted R2 computed with the observed and predicted values of blooming for each month. As the Fourier transform is highly flexible, it can adjust almost perfectly to the data: the median of weighted R2 was close to 1 (with a 95% confidence interval—from percentile 2.75 to 97.5—of 0.86.0 to 1).Figure 14Examples of observed time series of detections in cloud-free images (%) and their daily estimation modeled using the Fourier transform.Full size imageAfter the decomposition of the blooming signal, a daily time series of 1 year of ({widehat{bloom}}) was computed with the obtained parameters (365 values) (Fig. 14). Daily values of months with a weight of 0 were set to 0 as well as predicted negative values. Then all peaks and pits were identified in the ({widehat{bloom}}) time series. A peak or pit is an observation that is preceded and followed by, respectively, lower or higher observations62,63. For each peak, the day of start and stop were identified using the pit values. After this analysis, we were able to describe the blooming time series: that is, if there were one or more peaks and, for each peak, the days when the blooming initiated, peaked and stopped.To determine if different populations could be identified based on flowering timing, a cluster analysis was performed. A classical K-means clustering analysis was made on a dataset containing, for each pixel where Pleroma were detected, the days of start, peak, and end of blooming, the associated normalized blooming values and the xy coordinates of the pixels. If there were two peaks for a pixels, a line for each peak was created in the dataset. As the xy coordinates were in metres, they carried most of the variance in the dataset. To avoid the artefact of having clusters based only on the distance between pixels, the xy coordinates were divided by 100,000 and rounded to the nearest unit. Before the clustering analysis, all variable were scaled and centered. The number of clusters was determined based on the curve representing the total within-cluster sum of squares as a function of the number of clusters, and also to have the maximum number of clusters.To describe the association of Pleroma trees with environmental variables, we first reclassified each environmental variables into 10 classes according to the variable’s quantiles. Then a bootstrap procedure was applied. For the number N of Pleroma trees detections, N random point locations were sampled within the Atlantic Forest domain, and the value of each environmental variable at each point was extracted and stored. This operation was repeated 100 times. It enabled to compare the number of Pleroma trees in each quantile class with the mean and gave us a 95% confidence interval for the number of points obtained by random spatial sampling in each class. Using the elevation as an example, the null hypothesis of no spatial association between Pleroma trees and elevation was rejected at a level of 0.05% if the number of Pleroma trees in a quantile class of the elevation was outside the (0.025, 0.975) quantiles of the empirical distribution of elevation obtained by random location sampling in the same class. The same analysis of association with the environmental variables was made for the Handroanthus population identified by the K-means clustering analysis.All analyses were performed using R project software55. More

Magnesium soil contents and relationships to livestock healthMagnesium (Mg) deficiency (hypomagnesaemia) in ruminant livestock is a serious issue for the agricultural sector and accounts for a significant number of animal deaths annually. It is caused by a diet deficient in Mg, or due to an imbalance in the supply of Mg in comparison to other mineral cations1. Hypomagnesaemia is likely to be responsible for lower-productivity and diminished well-being in more animals in a herd compared to those displaying acute symptoms, given that herds/flocks generally receive a common diet2,3. If Mg deficiency could be prevented, it would be of benefit to both animal welfare and economic productivity. Recent research has confirmed that whilst hypomagnesaemia is commonly reported by UK farmers, the reported use of preventative measures is low, and the use of pasture interventions is lower still4. Pasture interventions can include the application of Mg-rich fertiliser or lime products, or selection of sward species with a propensity to take-up elevated Mg concentrations4,5.One aspect of dietary supply is the geographic control of pasture and farm-produced fodder. It is known that total Mg and plant-available Mg concentrations in soil are controlled by geological and geographic factors6,7,8,9 and that there is little evidence for any changes in pasture soil Mg concentrations through time6,10. The magnesium content of soil relates to that of the bedrock, where it is high in the bedrock it is high in soil and vice versa. Thus, the composition of all pasture and farm-grown fodder will be influenced by this natural environmental endowment as well as pasture management decisions.Grass productivity and soil pHA key pasture management activity is that of soil pH—which here is reported as measured in water (pHw), consistent with standard agronomic laboratory practice in the UK11. Grassland mineral soil is recommended to be maintained at pHw ≥ 6.0 in Britain12,13. In Ireland, where grass-clover pasture is more widely practiced, a pHw threshold of 6.5 is recommended14. However, multiple lines of evidence exist that indicate pasture soil is frequently below these pHw recommendations. Private sector on-farm sample data summaries from the UK consistently show pH typically below recommendations in pasture soil: the most recent annual data synthesis reports 57% of grassland soil with pHw ≤ 5.99, and 27% with pHw 6.00–6.498. This is consistent with systematically collected public sector data across the north of Ireland, where 84% of pasture samples were below the clover-grass recommended threshold of pHw 6.515.Grassland production is widespread in Wales and western England (Fig. 1). Two environmental factors jointly contributing to the lower pH in these regions are (1) geological—these areas are most often on soils which are developed over rocks with low concentrations of base cations (Figs. 1 and 2); and, (2) these areas are also often upland areas, associated with typically higher rainfall16 which will further leach base cations. Added to these environmental factors are the application of nitrogen fertilisers which have an acidifying effect17. Thus, many pastoral areas require treatment using agricultural lime in order to optimize soil pH for grass growth18.The use of liming materialsThe opportunity to improve grazing livestock Mg nutrition through use of Mg-rich lime is identified in guidance available to farmers12. This can have the dual benefits of maintaining soil pH for grass growth and ensuring Mg levels in livestock feed is at sufficient levels19,20,21. The combination of soil treatment for pH and Mg would therefore appear to be an efficient solution to solve issues surrounding Mg deficiency22. Conversely, for many soils with existing high Mg levels it may be important to treat with low Mg liming materials to ensure an optimal Ca–Mg balance to preserve the soil structure23.The use of Mg lime is only one of many methods of controlling Mg levels in livestock feed. Other methods, such as direct additions to feeds, salt licks, pelletised fertilisers products are also effective in reducing incidences of hypomagnesaemia and need to be considered as part of holistic review of a individual farms requirements, this is discussed in Kumssa et al.5.Maintaining optimal soil pH will directly affect the productivity of grass used for grazing, and will increase fertiliser use efficiency14. However, in some cases, for example upland sheep farming, a low investment—low return approach, with minimum interventions such as liming, may be entirety sensible and appropriate to the farm business and local landscape24,25.The extent to which agricultural lime is used in Britain is captured through the annual British Survey of Fertiliser Practice (BSFP) and can also be inferred from commodity production statistics. Production can be regarded as a good proxy for consumption since due to its high bulk and low price it is not exported in significant quantities.The BSFP, is an annual Department for Environment, Food and Rural Affairs (DEFRA) survey26, which representatively samples fertiliser and lime use across the British farming sector. This captures information on lime use in three geological material categories as used in arable and pastoral systems, as well as use of sugar beet lime and ‘other’ options. Sugar beet lime use is very low on grassland (generally unrecorded on ‘permanent’ pasture); ‘other’ categories are generally on a par with Mg-lime, but more detailed liming characteristics are not reported. Figure 3 shows a clear trend in decreasing production of agricultural liming material over the last 40 years. Lime use in the UK peaked in the late 1950s and mid 1960s likely due to a subsidy for agricultural lime in place at the time, this ended in 1978, causing prices to increase and subsequently lime use to decrease27. The use of agricultural lime has continued on a declining, or flat trend, likely due to reluctance to engage in soil treatments that are seen to be costly and a lack of knowledge over its potential benefits.Figure 1Classes of land cover considered by this study, based on Land Cover Map 201533. Some features of this map are based on digital spatial data licensed from the UK Centre for Ecology and Hydrology. Created using ArcMap 10.7.1, ESRI, 2019.Full size imageFigure 4 shows the use of geological lime products to be low at present in respect of the proportion of fields to which lime is reported to be applied, and that this is particularly pervasive on permanent grassland, with a 10-year average to 2019 of 2.9% (range 2.0–4.1%). Of this, a 10-year average of 0.4% of fields had Mg-lime applied, with 1.8% of fields having limestone applied. The limited use of chalk (0.1% of fields) probably reflects the distance between the majority of pasture and the outcrop of the chalk. Recent grassland ( More

1.de Wet, J. M. J. & Harlan, J. R. Weeds and domesticates: Evolution in the man-made habitat. Econ. Bot. 29(2), 99–108. https://doi.org/10.1007/BF02863309 (1975).Article 

Google Scholar 
2.Ceballos, G. et al. Accelerated modern human—Induced species losses: Entering the sixth mass extinction. Sci. Adv. 1(June), 1–6. https://doi.org/10.1126/sciadv.1400253 (2015).Article 

Google Scholar 
3.Wilcove, D. S. Nest predation in forest tracts and the decline of migratory songbirds. Ecology 66(4), 1211–1214 (1985).Article 

Google Scholar 
4.Airoldi, L. & Bulleri, F. Anthropogenic disturbance can determine the magnitude of opportunistic species responses on marine urban infrastructures. PLoS ONE https://doi.org/10.1371/journal.pone.0022985 (2011).Article 
PubMed 
PubMed Central 

Google Scholar 
5.Baker, H. G. The evolution of weeds. Annu. Rev. Ecol. Syst. 5, 1–24. https://doi.org/10.2307/2096877 (1974).ADS 
Article 

Google Scholar 
6.Richardson, D. M. et al. Naturalization and invasion of alien plants: Concepts and definitions. Divers. Distrib. 6, 93–107 (2008).Article 

Google Scholar 
7.van Etten, M. L., Conner, J. K., Chang, S. M. & Baucom, R. S. Not all weeds are created equal: A database approach uncovers differences in the sexual system of native and introduced weeds. Ecol. Evol. 7(8), 2636–2642. https://doi.org/10.1002/ece3.2820 (2017).Article 
PubMed 
PubMed Central 

Google Scholar 
8.Booth, B. D. & Swanton, C. J. Assembly theory applied to weed communities 50th Anniversary—Invited Article Assembly theory applied to weed communities. Weed Sci. 50(3), 2–13. https://doi.org/10.1614/0043-1745(2002)050 (2002).CAS 
Article 

Google Scholar 
9.Kuester, A., Conner, J. K., Culley, T. & Baucom, R. S. How weeds emerge: A taxonomic and trait-based examination using United States data. New Phytol. 202(3), 1055–1068. https://doi.org/10.1111/nph.12698 (2014).Article 
PubMed 
PubMed Central 

Google Scholar 
10.van Kleunen, M. et al. The ecology and evolution of alien plants. Annu. Rev. Ecol. Evol. Syst. https://doi.org/10.1146/annurev-ecolsys-110617-062654 (2018).Article 

Google Scholar 
11.de Bona, S. et al. Spatio-temporal dynamics of density-dependent dispersal during a population colonisation. Ecol. Lett. 22, 634–644 (2019).PubMed 
Article 

Google Scholar 
12.Baker, H. G. Self-compatibility and establishment after “long-distance” dispersal. Evolution 9(3), 347. https://doi.org/10.2307/2405656 (1955).Article 

Google Scholar 
13.Razanajatovo, M. et al. Plants capable of selfing are more likely to become naturalized. Nat. Commun. 7, 13313. https://doi.org/10.1038/ncomms13313 (2016).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
14.Vallejo-Marín, M., Dorken, M. E. & Barrett, S. C. H. The ecological and evolutionary consequences of clonality for plant mating. Annu. Rev. Ecol. Evol. Syst. 41(1), 193–213. https://doi.org/10.1146/annurev.ecolsys.110308.120258 (2010).Article 

Google Scholar 
15.Rodger, J. G., Van Kleunen, M. & Johnson, S. D. Pollinators, mates and Allee effects: The importance of self-pollination for fecundity in an invasive lily. Funct. Ecol. 27(4), 1023–1033. https://doi.org/10.1111/1365-2435.12093 (2013).Article 

Google Scholar 
16.Barrett, S. C. H. & Harder, L. D. The ecology of mating and its evolutionary consequences in seed plants. Annu. Rev. Ecol. Evol. Syst. https://doi.org/10.1146/annurev-ecolsys-110316-023021 (2017).Article 

Google Scholar 
17.Klimeš, L., Klimešová, J., Hendriks, R. & van Groenendael, J. Clonal plant architecture: A comparative analysis of form and function. In The Ecology and Evolution of Clonal Plants (eds De Kroon, H. & Van Groenendael, J. M.) 1–29 (Backhuys, 1997).
Google Scholar 
18.Barrett, S. C. H. Influences of clonality on plant sexual reproduction. Proc. Natl. Acad. Sci. 112(29), 8859–8866. https://doi.org/10.1073/pnas.1501712112 (2015).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
19.Heywood, J. S. Spatial analysis of genetic variation in plant populations. Annu. Rev. Ecol. Syst. 22, 335–355 (1991).Article 

Google Scholar 
20.Barrett, S. C. H. Evolution of mating systems: Outcrossing versus selfing. In The Princeton Guide to Evolution (ed. Losos, J. B.) 356–362 (Princeton University Press, 2013).
Google Scholar 
21.Barrett, S. C. H., Arunkumar, R. & Wright, S. I. The demography and population genomics of evolutionary transitions to self-fertilization in plants. Philos. Trans. R. Soc. B Biol. Sci. 369(1648), 20130344 (2014).Article 

Google Scholar 
22.Picó, F. X., Quintana-Ascencio, P. F., Mildén, M., Ehrlén, J. & Pfingsten, I. Modelling the effects of genetics and habitat on the demography of a grassland herb. Basic Appl. Ecol. 10(2), 122–130. https://doi.org/10.1016/j.baae.2008.02.006 (2009).Article 

Google Scholar 
23.Ellstrand, N. C. & Roose, M. L. Patterns of genotypic diversity in clonal plant species. Am. J. Bot. 74, 123–131 (1987).Article 

Google Scholar 
24.Loh, R., Scarano, F. R., Alves-Ferreira, M. & Salgueiro, F. Implications of clonality to population genetic structure of the nurse species Aechmea nuducaulis (L.) Griseb. (Bromeliaceae). Bot. J. Linn. Soc. 178, 329–341 (2015).Article 

Google Scholar 
25.Hedrick, P. W. Purging inbreeding depression and the probability of extinction: Full-sib mating. Heredity 73, 363–372. https://doi.org/10.1038/hdy.1994.183 (1994).Article 
PubMed 

Google Scholar 
26.Arunkumar, R., Ness, R. W., Wright, S. I. & Barrett, S. C. H. The evolution of selfing is accompanied by reduced efficacy of selection and purging of deleterious mutations. Genetics 199(3), 817–829. https://doi.org/10.1534/genetics.114.172809 (2015).Article 
PubMed 

Google Scholar 
27.Pannell, J. R. & Barrett, S. C. H. Baker’s law revisited: Reproductive assurance in a metapopulation. Evolution 52(3), 657–668. https://doi.org/10.2307/2411261 (1998).Article 
PubMed 

Google Scholar 
28.Hamrick, J. L. & Trapnell, D. W. Using population genetic analyses to understand seed dispersal patterns. Acta Oecologica 37, 641–649 (2011).ADS 
Article 

Google Scholar 
29.Côrtes, M. C. et al. Low plant density enhances gene dispersal in the Amazonian understory herb Heliconia acuminata. Mol. Ecol. 22, 5716–5729 (2013).PubMed 
Article 
CAS 

Google Scholar 
30.Trapnell, D. W., Hamrick, J. L., Ishibashi, C. D. & Kartzinel, T. R. Genetic inference of epiphytic orchid colonization; it may only take one. Mol. Ecol. 22, 3680–3692. https://doi.org/10.1111/mec.12338 (2013).Article 
PubMed 

Google Scholar 
31.Chung, M. Y. et al. Fine-scale genetic structure in populations of the spring ephemeral herb Megaleranthis saniculifolia (Ranunculaceae). Flora Morphol. Distrib. Funct. Ecol. Plants 240, 16–24 (2018).
Google Scholar 
32.Roberts, N. R., Dalton, P. J. & Jordan, G. J. Epiphytic ferns and bryophytes of Tasmanian tree-ferns: A comparison of diversity and composition between two host species. Austral Ecol. 30(2), 146–154. https://doi.org/10.1111/j.1442-9993.2005.01440.x (2005).Article 

Google Scholar 
33.Cardelús, C. L. & Chazdon, R. L. Inner-crown microenvironments of two emergent tree species in a lowland wet forest. Biotropica 37(2), 238–244. https://doi.org/10.1111/j.1744-7429.2005.00032.x (2005).Article 

Google Scholar 
34.Quaresma, A. C., Piedade, M. T. F., Wittmann, F. & ter Steege, H. Species richness, composition, and spatial distribution of vascular epiphytes in Amazonian black-water floodplain forests. Biodivers. Conserv. 27(8), 1981–2002. https://doi.org/10.1007/s10531-018-1520-3 (2018).Article 

Google Scholar 
35.Claver, F. K., Alaniz, J. R. & Caldíz, D. O. Tillandsia spp.: Epiphytic weeds of trees and bushes. For. Ecol. Manag. 6(4), 367–372. https://doi.org/10.1016/0378-1127(83)90044-0 (1983).Article 

Google Scholar 
36.Bartoli, C. G., Beltrano, J., Fernández, L. V. & Caldíz, D. O. Control of the epiphytic weeds Tillandsia recurvata and Tillandsia aeranthos with different herbicides. For. Ecol. Manage. 59, 289–294 (1993).Article 

Google Scholar 
37.Flores-Palacios, A., García-Franco, J. G. & Capistrán-Barradas, A. Biomass, phorophyte specificity and distribution of Tillandsia recurvata in a tropical semi-desert environment (Chihuahuan Desert, Mexico). Plant Ecology and Evolution 148(1), 68–75 (2015).Article 

Google Scholar 
38.Birge, W. I. The anatomy and some biological aspects of the “ball moss”, Tillandsia recurvata, 1–24. L. Bull. Univ. Tex. 194(20) (1911).39.Smith, L. B. & Downs, R. J. Tillandsioideae (Bromeliaceae). In Flora Neotropica Monograph 14(2), 663–1492 (1977).40.Hewitt, G. M. (1996). Some genetic consequences of ice ages, and their role in speciation. Biological Journal of the Linnaean Society, 58(July), 247–276. Retrieved from papers3://publication/uuid/B9DB7D5E-D6AE-404C-BFFC-9F813345329441.McWilliams, E. Chronology of the Natural Range Expansion of Tillandsia recurvata (Bromeliaceae) in Texas. Contributions to Botany 15(2), 343–346 (1992).42.Flores-Palacios, A., Barbosa-Duchateau, C. L., Valencia-Díaz, S., Capistrán-Barradas, A. & García-Franco, J. G. Direct and indirect effects of Tillandsia recurvata on Prosopis laevigata in the Chihuahua desert scrubland of San Luis Potosi, Mexico. J. Arid Environ. 104, 88–95. https://doi.org/10.1016/j.jaridenv.2014.02.010 (2014).ADS 
Article 

Google Scholar 
43.Benzing, D. H. Bromeliaceae: Profile of an Adaptive Radiation (Cambridge University Press, 2000).Book 

Google Scholar 
44.Benzing, D. H. Air Plants: Epiphytes and Aerial Gardens (Cornell University Press, 2012).Book 

Google Scholar 
45.Foster, M. D. Blueprint of the jungle as depicted by the altitude of growth of the Bromeliadswith notes on the culture of certain tropical epiphytes. Bull. N. Y. Bot. Garden 46, 9–16 (1945).
Google Scholar 
46.Soltis, D. E., Gilmartin, A. J., Rieseberg, L. & Gardner, S. Genetic variation in the epiphytes Tillandsia ionantha and T. recurvata (Bromeliaceae). Am. J. Bot. 74(4), 531–537 (1987).CAS 
Article 

Google Scholar 
47.Orozco-Ibarrola, O. A., Flores-Hernández, P. S., Victoriano-Romero, E., Corona-López, A. M. & Flores-Palacios, A. Are breeding system and florivory associated with the abundance of Tillandsia species (Bromeliaceae)?. Bot. J. Linn. Soc. 177(1), 50–65. https://doi.org/10.1111/boj.12225 (2015).Article 

Google Scholar 
48.Chilpa-Galván, N. et al. Seed traits favouring dispersal and establishment of six epiphytic Tillandsia (Bromeliaceae) species. Seed Sci. Res. https://doi.org/10.1017/S0960258518000247 (2018).Article 

Google Scholar 
49.Southwood, T. & Kennedy, C. Trees as islands. Oikos 41(3), 359–371. https://doi.org/10.2307/3544094 (1983).Article 

Google Scholar 
50.Burns, K. C. Network properties of an epiphyte metacommunity. J. Ecol. 95(5), 1142–1151 (2007).Article 

Google Scholar 
51.Trapnell, D. W., Hamrick, J. L. & Nason, J. D. Three-dimensional fine-scale genetic structure of the neotropical epiphytic orchid, Laelia rubescens. Mol. Ecol. 13, 1111–1118 (2004).CAS 
PubMed 
Article 

Google Scholar 
52.Torres, E., Riofrío, M.-L. & Iriondo, J. M. Complex fine-scale spatial genetic structure in Epidendrum rhopalostele: an epiphytic orchid. Heredity https://doi.org/10.1038/s41437-018-0139-1 (2018).Article 
PubMed 
PubMed Central 

Google Scholar 
53.Victoriano-Romero, E., Valencia-Díaz, A., Toledo-Hernández, V. H. & Flores-Palacios, A. Dispersal limitation of Tillandsia species correlates with rain and host structure in a central Mexican tropical dry forest. PLoS ONE 12(2), e0171614 (2017).PubMed 
PubMed Central 
Article 
CAS 

Google Scholar 
54.Martins, S. E. (2009). Flora fanerogâmica do estado de São Paulo. FAPESP: Instituto de Botânica.55.Chaves, C. J. N., Dyonisio, J. C. J. C. & Rossatto, D. R. D. R. Host trait combinations drive abundance and canopy distribution of atmospheric bromeliad assemblages. AoB Plants 8(October 2015), plw010. https://doi.org/10.1093/aobpla/plw010 (2016).Article 
PubMed 
PubMed Central 

Google Scholar 
56.Epps, C. W. & Keyghobadi, N. Landscape genetics in a changing world: Disentangling historical and contemporary influences and inferring change. Mol. Ecol. 24(24), 6021–6040. https://doi.org/10.1111/mec.13454 (2015).Article 
PubMed 

Google Scholar 
57.Cushman, S. A., Shirk, A. & Landguth, E. L. Separating the effects of habitat area, fragmentation and matrix resistance on genetic differentiation in complex landscapes. Landsc. Ecol. 27(3), 369–380. https://doi.org/10.1007/s10980-011-9693-0 (2012).Article 

Google Scholar 
58.Jackson, N. D. & Fahrig, L. Habitat amount, not habitat configuration, best predicts population genetic structure in fragmented landscapes. Landsc. Ecol. 31(5), 951–968. https://doi.org/10.1007/s10980-015-0313-2 (2016).Article 

Google Scholar 
59.Grimm, V. & Railsback, S. F. Individual-Based Modelling and Ecology (Princeton University Press, 2005).MATH 
Book 

Google Scholar 
60.Csilléry, K., Blum, M. G. B., Gaggiotti, O. E. & François, O. Approximate Bayesian Computation (ABC) in practice. Trends Ecol. Evol. 25(7), 410–418. https://doi.org/10.1016/j.tree.2010.04.001 (2010).Article 
PubMed 

Google Scholar 
61.Udupa, S. M. & Baum, M. High mutation rate and mutational bias at (TAA)n microsatellite loci in chickpea (Cicer arietinum L.). Mol. Genet. Genom. 265(6), 1097–1103. https://doi.org/10.1007/s004380100508 (2001).CAS 
Article 

Google Scholar 
62.Anmarkrud, J. A., Kleven, O., Bachmann, L. & Lifjeld, J. T. Microsatellite evolution: Mutations, sequence variation, and homoplasy in the hypervariable avian microsatellite locus HrU10. BMC Evol. Biol. 8(1), 1–10. https://doi.org/10.1186/1471-2148-8-138 (2008).CAS 
Article 

Google Scholar 
63.Marriage, T. N. et al. Direct estimation of the mutation rate at dinucleotide microsatellite loci in Arabidopsis thaliana (Brassicaceae). Heredity 103(4), 310–317. https://doi.org/10.1038/hdy.2009.67 (2009).CAS 
Article 
PubMed 

Google Scholar 
64.Bernal, R., Valverde, T. & Hernández-Rosas, L. Habitat preference of the epiphyte Tillandsia recurvata (Bromeliaceae) in a semi-desert environment in Central Mexico. Can. J. Bot. 83(10), 1238–1247 (2005).Article 

Google Scholar 
65.Chaves, C. J. & Rossatto, D. R. Unravelling intricate interactions among atmospheric bromeliads with highly overlapping niches in seasonal systems. Plant Biol. 22(2), 243–251 (2020).CAS 
PubMed 
Article 

Google Scholar 
66.Vekemans, X. & Hardy, O. J. New insights from fine-scale spatial genetic structure analyses in plant populations. Mol. Ecol. 13(4), 921–935. https://doi.org/10.1046/j.1365-294X.2004.02076.x (2004).CAS 
Article 
PubMed 

Google Scholar 
67.Ward, S. Genetic analysis of invasive plant populations at different spatial scales. Biol. Invasions 8(3), 541–552. https://doi.org/10.1007/s10530-005-6443-8 (2006).Article 

Google Scholar 
68.Pettengill, J. B., Briscoe Runquist, R. D. & Moeller, D. A. Mating system divergence affects the distribution of sequence diversity within and among populations of recently diverged subspecies of Clarkia xantiana (Onagraceae). Am. J. Bot. 103(1), 99–109. https://doi.org/10.3732/ajb.1500147 (2016).CAS 
Article 
PubMed 

Google Scholar 
69.Atwater, D. Z., Fletcher, R. A., Dickinson, C. C., Paterson, A. H. & Barney, J. N. Evidence for fine-scale habitat specialization in an invasive weed. J. Plant Ecol. 11(2), 189–199. https://doi.org/10.1093/jpe/rtw124 (2018).Article 

Google Scholar 
70.Li, J. & Dong, M. Fine-scale clonal structure and diversity of invasive plant Mikania micrantha H.B.K. and its plant parasite Cuscuta campestris Yunker. Biol. Invasions 11(3), 687–695. https://doi.org/10.1007/s10530-008-9283-5 (2009).MathSciNet 
Article 

Google Scholar 
71.Ren, M. X., Cafasso, D., Cozzolino, S. & Pinheiro, F. Extensive genetic differentiation at a small geographical scale: Reduced seed dispersal in a narrow endemic marsh orchid, Anacamptis robusta. Bot. J. Linn. Soc. 183(3), 429–438. https://doi.org/10.1093/botlinnean/bow017 (2017).Article 

Google Scholar 
72.Barluenga, M. et al. Fine-scale spatial genetic structure and gene dispersal in Silene latifolia. Heredity 106(1), 13–24. https://doi.org/10.1038/hdy.2010.38 (2011).CAS 
Article 
PubMed 

Google Scholar 
73.Charbonneau, A. et al. Weed evolution: Genetic differentiation among wild, weedy, and crop radish. Evol. Appl. https://doi.org/10.1111/eva.12699 (2018).Article 
PubMed 
PubMed Central 

Google Scholar 
74.Sagnard, F., Oddou-Muratorio, S., Pichot, C., Vendramin, G. G. & Fady, B. Effects of seed dispersal, adult tree and seedling density on the spatial genetic structure of regeneration at fine temporal and spatial scales. Tree Genet. Genomes 7(1), 37–48. https://doi.org/10.1007/s11295-010-0313-y (2011).Article 

Google Scholar 
75.Counsens, R. & Mortimer, M. Dynamics of Weed Populations (Cambridge University Press, 1995).Book 

Google Scholar 
76.Loreau, M. et al. Unifying sources and sinks in ecology and Earth sciences. Biol. Rev. 88, 365–379 (2013).PubMed 
Article 

Google Scholar 
77.dos Santos, L. S. et al. Generalized Allee effect model. Theory Biosci. 133, 117–124 (2014).PubMed 

Google Scholar 
78.Spruch, L. et al. Modeling community assembly on growing habitat “islands”: A case study on trees and their vascular epiphyte communities. Theor. Ecol. 12, 1–17 (2019).Article 

Google Scholar 
79.Einzmann, H. J. R. & Zotz, G. “No signs of saturation”: long-term dynamics of vascular epiphyte communities in a human-modified landscape. Biodivers. Conserv. 26, 1393–1410 (2017).Article 

Google Scholar 
80.Belinchón, R., Harrison, P. J., Mair, L., Várkonyi, G. & Snäll, T. Local epiphyte establishment and future metapopulation dynamics in landscapes with different spatiotemporal properties. Ecology 98(3), 741–750. https://doi.org/10.1002/ecy.1686 (2017).Article 
PubMed 

Google Scholar 
81.Vergara-Torres, C. A., Pacheco-Álvarez, M. C. & Flores-Palacios, A. Host preference and host limitation of vascular epiphytes in a tropical dry forest of central Mexico. J. Trop. Ecol. 26(6), 563–570. https://doi.org/10.1017/S0266467410000349 (2010).Article 

Google Scholar 
82.Barrett, S. C. H. & Kohn, J. R. Genetic and evolutionary consequences of small population size in plants: Implications for conservation. In Genetics and Conservation of Rare Plants (eds Falk, D. A. & Holsinge, K. E.) 3–30 (Oxford University Press, 1991).
Google Scholar 
83.Nathan, R., Horn, H. S., Chave, J. & Levin, S. A. Mechanistic models for tree seed dispersal by wind in dense forests and open landscapes. In Seed Dispersal and Frugivory-Ecologie, Evolution, Conservation 69–82 (2002). https://doi.org/10.1079/9780851995250.006984.Cousens, R. et al. Dispersal in Plants. A Population Perspective (Oxford University Press, 2008).Book 

Google Scholar 
85.Snäll, T., Ehrlén, J. & Rydin, H. Colonization-extinction dynamics of an epiphyte metapopulation in a dynamic landscape. Ecology 86(1), 106–115 (2005).Article 

Google Scholar 
86.Ruiz-Cordova, J. P., Toledo-Hernández, V. H. & Flores-Palacios, A. The effect of substrate abundance in the vertical stratification of bromeliad epiphytes in a tropical dry forest (Mexico). Flora Morphol. Distrib. Funct. Ecol. Plants 209(8), 375–384. https://doi.org/10.1016/j.flora.2014.06.003 (2014).Article 

Google Scholar 
87.Flores-Palacios, A., Bustamante-Molina, A. B., Corona-López, A. M. & Valencia-Díaz, S. Seed number, germination and longevity in wild dry forest Tillandsia species of horticultural value. Scientia Hortic. 187, 72–79 (2015).Article 

Google Scholar 
88.Goodman, R., & Herold, M. (2014). Why maintaining tropical forests is essential and urgent for a stable climate. Center for Global Development Working Paper, (385).89.Seymour, F. & Busch, J. Why Forests? Why Now?: The Science, Economics, and Politics of Tropical Forests and Climate Change (Brookings Institution Press, 2016).
Google Scholar 
90.Stephenson, N. L. et al. Rate of tree carbon accumulation increases continuously with tree size. Nature 507(7490), 90–93 (2014).ADS 
CAS 
PubMed 
Article 

Google Scholar 
91.Tel-Zur, N., Abbo, S., Myslabodsky, D. & Mizrahi, Y. Modified CTAB procedure for DNA isolation from epiphytic cacti of genera Hylocereus and Selenicereus (Cactaceae). Plant Mol. Biol. Rep. 17, 249–254 (1999).CAS 
Article 

Google Scholar 
92.Chaves, C. J. N., Aoki-Gonçalves, F., Leal, B. S. S., Rossatto, D. R. & Palma-Silva, C. Transferability of nuclear microsatellite markers to the atmospheric bromeliads Tillandsia recurvata and T. aeranthos (Bromeliaceae). Braz. J. Bot. 41, 931–935. https://doi.org/10.1007/s40415-018-0494-4 (2018).Article 

Google Scholar 
93.Keenan, K., Mcginnity, P., Cross, T. F., Crozier, W. W. & Prodöhl, P. A. DiveRsity: An R package for the estimation and exploration of population genetics parameters and their associated errors. Methods Ecol. Evol. https://doi.org/10.1111/2041-210X.12067 (2013).Article 

Google Scholar 
94.Slatkin, M. A measure of population subdivision based on microsatellite allele frequencies. Genetics 139, 457–462 (1995).CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
95.Nei, M. Genetic distances between populations. Am. Nat. 106, 283–292 (1972).Article 

Google Scholar 
96.Edwards, A. W. F. Distance between populations on the basis of gene frequencies. Biometrics 27, 873–881 (1971).CAS 
PubMed 
Article 

Google Scholar 
97.Reynolds, J. B., Weir, B. S. & Cockerham, C. C. Estimation of the coancestry coefficient: Basis for a short-term genetic distance. Genetics 105, 767–779 (1983).CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
98.Kamvar, Z. N., Tabima, J. F. & Grünwald, N. J. Poppr: An R package for genetic analysis of populations with clonal, partially clonal, and/or sexual reproduction. PeerJ https://doi.org/10.7717/peerj.281 (2014).Article 
PubMed 
PubMed Central 

Google Scholar 
99.Paradis, E. pegas: an R package for population genetics with an integrated-modular approach. Bioinformatics 26, 419–420 (2010).CAS 
PubMed 
Article 

Google Scholar 
100.Excoffier, L., Smouse, P. E. & Quattro, J. M. Analysis of molecular variance inferred from metric distances among DNA haplotypes: Application to human mitochondrial DNA restriction data. Genetics 131, 479–491. https://doi.org/10.1007/s00424-009-0730-7 (1992).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
101.Loiselle, B. A., Sork, V. L., Nason, J. & Graham, C. Spatial genetic structure of a tropical understory shrub, Psychotria officinalis (Rubiaceae). Am. J. Bot. 82(11), 1420–1425 (1995).Article 

Google Scholar 
102.Bailleul, D., Stoeckel, S. & Arnaud-Haond, S. RClone: A package to identify MultiLocus Clonal Lineages and handle clonal data sets in r. Methods Ecol. Evol. 7(8), 966–970. https://doi.org/10.1111/2041-210X.12550 (2016).Article 

Google Scholar 
103.Harrison, S. et al. Beta diversity on geographic gradients in Britain. J. Anim. Ecol. 61(1), 151–158 (1992).Article 

Google Scholar 
104.Jost, L. Partitioning diversity into independent alpha and beta components. Ecology 88(10), 2427–2439. https://doi.org/10.1890/07-1861.1 (2007).Article 
PubMed 

Google Scholar 
105.Charney, N. & Record, S. Vegetarian: Jost diversity measures for community data. https://cran.r-project.org/web/packages/vegetarian/index.html (2012). Accessed Jul 2018.106.Wilensky, U. NetLogo (Northwestern University, Center for Connected Learning and Computer-Based Modeling, 1999).
Google Scholar 
107.Grimm, V. et al. A standard protocol for describing individual-based and agent-based models. Ecol. Model. 198(1–2), 115–126. https://doi.org/10.1016/j.ecolmodel.2006.04.023 (2006).Article 

Google Scholar 
108.Grimm, V. et al. The ODD protocol: A review and first update. Ecol. Model. 221(23), 2760–2768. https://doi.org/10.1016/j.ecolmodel.2010.08.019 (2010).Article 

Google Scholar 
109.Kooijman, B. & Kooijman, S. A. L. M. Dynamic Energy Budget Theory for Metabolic Organisation (Cambridge University Press, 2010).
Google Scholar 
110.Sibly, R. M. et al. Representing the acquisition and use of energy by individuals in agent-based models of animal populations. Methods Ecol. Evol. 4(2), 151–161 (2013).Article 

Google Scholar 
111.Johnston, A. S. A., Hodson, M. E., Thorbek, P., Alvarez, T. & Sibly, R. M. An energy budget agent-based model of earthworm populations and its application to study the effects of pesticides. Ecol. Model. 280, 5–17 (2014).CAS 
Article 

Google Scholar 
112.van der Vaart, E., Johnston, A. S. A. & Sibly, R. M. Predicting how many animals will be where: How to build, calibrate and evaluate individual-based models. Ecol. Model. 326, 113–123 (2016).Article 

Google Scholar 
113.Garza, J. C. & Williamson, E. G. Detection of reduction in population size using data from microsatellite loci. Mol. Ecol. 10, 305–318 (2001).CAS 
PubMed 
Article 

Google Scholar 
114.Excoffier, L., Laval, G. & Schneider, S. Arlequin (version 3.0): An integrated software package for population genetics data analysis. Evol. Bioinform. Online 1, 47–50 (2005).CAS 
Article 

Google Scholar 
115.Csilléry, K., François, O. & Blum, M. G. abc: An R package for approximate Bayesian computation (ABC). Methods Ecol. Evol. 3(3), 475–479 (2012).Article 

Google Scholar 
116.Pastur, G. M., Lencinas, M. V., Cellini, J. M. & Mundo, I. Diameter growth: Can live trees decrease?. Forestry 80(1), 83–88. https://doi.org/10.1093/forestry/cpl047 (2007).Article 

Google Scholar  More

1.Rubel, F. et al. Geographical distribution of Dermacentor marginatus and Dermacentor reticulatus in Europe. Ticks Tick Borne Dis. 7, 224–233 (2016).PubMed 
Article 
PubMed Central 

Google Scholar 
2.Medlock, J. M. et al. Driving forces for changes in geographical distribution of Ixodes ricinus ticks in Europe. Parasites Vectors 6, 1–11 (2013).PubMed 
PubMed Central 
Article 

Google Scholar 
3.Jongejan, F. & Uilenberg, G. The global importance of ticks. Parasitology 129, 3–14 (2004).Article 

Google Scholar 
4.Földvári, G., Široký, P., Szekeres, S., Majoros, G. & Sprong, H. Dermacentor reticulatus: a vector on the rise. Parasites Vectors 9, 1–29 (2016).Article 

Google Scholar 
5.Ličková, M. et al. Dermacentor reticulatus is a vector of tick-borne encephalitis virus. Ticks Tick Borne Dis. 11, 101414 (2020).PubMed 
Article 
PubMed Central 

Google Scholar 
6.Pawełczyk, A. et al. Long-term study of Borrelia and Babesia prevalence and co-infection in Ixodes ricinus and Dermacentor recticulatus ticks removed from humans in Poland, 2016–2019. Parasites Vectors 14, 1–13 (2021).Article 
CAS 

Google Scholar 
7.Karbowiak, G. et al. The competition between immatures of Ixodes ricinus and Dermacentor reticulatus (Ixodida: Ixodidae) ticks for rodent hosts. J. Med. Entomol. 56, 448–452 (2018).Article 

Google Scholar 
8.Karbowiak, G. The occurrence of the Dermacentor reticulatus tick-its expansion to new areas and possible causes. Ann. Parasitol. 60, 37–47 (2014).PubMed 
PubMed Central 

Google Scholar 
9.Drehmann, M. et al. The Spatial Distribution of Dermacentor Ticks (Ixodidae) in Germany: Evidence of a continuing spread of Dermacentor reticulatus. Front. Vet. Sci. 7, 578220 (2020).PubMed 
PubMed Central 
Article 

Google Scholar 
10.Sands, B. O., Bryer, K. E. & Wall, R. Climate and the seasonal abundance of the tick Dermacentor reticulatus. Med. Vet. Entomol. https://doi.org/10.1111/mve.12518 (2021).Article 
PubMed 

Google Scholar 
11.Hasle, G. et al. Transport of ticks by migratory passerine birds to Norway. J. Parasitol. 95, 1342–1351 (2009).PubMed 
Article 
PubMed Central 

Google Scholar 
12.Kjær, L. J. et al. A large-scale screening for the taiga tick, Ixodes persulcatus, and the meadow tick, Dermacentor reticulatus, in southern Scandinavia, 2016. Parasites Vectors 12, 1–4 (2019).Article 

Google Scholar 
13.García-Sanmartín, J., Barandika, J. F., Juste, R. A., García-Pérez, A. L. & Hurtado, A. Distribution and molecular detection of Theileria and Babesia in questing ticks from northern Spain. Med. Vet. Entomol. 22, 318–325 (2008).PubMed 
Article 
PubMed Central 

Google Scholar 
14.Olivieri, E. et al. The southernmost foci of Dermacentor reticulatus in Italy and associated Babesia canis infection in dogs. Parasites Vectors 9, 1–9 (2016).Article 
CAS 

Google Scholar 
15.Široký, P. et al. The distribution and spreading pattern of Dermacentor reticulatus over its threshold area in the Czech Republic: How much is range of this vector expanding?. Vet. Parasitol. 183, 130–135 (2011).PubMed 
Article 
PubMed Central 

Google Scholar 
16.Hornok, S. & Farkas, R. Influence of biotope on the distribution and peak activity of questing ixodid ticks in Hungary. Med. Vet. Entomol. 23, 41–46 (2009).CAS 
PubMed 
Article 
PubMed Central 

Google Scholar 
17.Chitimia-Dobler, L. Spatial distribution of Dermacentor reticulatus in Romania. Vet. Parasitol. 214, 219–223 (2015).PubMed 
Article 

Google Scholar 
18.Akimov, I. & Nebogatkin, I. Distribution of Ticks from of the Genus Dermacentor (Acari, Ixodidae) in Ukraine. Vestnik Zoologii 45, 6 (2011).
Google Scholar 
19.Kiewra, D., Szymanowski, M., Czułowska, A. & Kolanek, A. The local-scale expansion of Dermacentor reticulatus ticks in Lower Silesia, SW, Poland. Ticks Tick Borne Dis. 12, 101599 (2021).PubMed 
Article 

Google Scholar 
20.Dwużnik-Szarek, D. et al. Monitoring the expansion of Dermacentor reticulatus and occurrence of canine babesiosis in Poland in 2016–2018. Parasites Vectors 14, 1–18 (2021).Article 

Google Scholar 
21.Zając, Z., Woźniak, A. & Kulisz, J. Density of Dermacentor reticulatus ticks in eastern Poland. Int. J. Environ. Res. Public Health 17, 2814 (2020).PubMed Central 
Article 

Google Scholar 
22.Ogden, N. H., Ben Beard, C., Ginsberg, H. S. & Tsao, J. I. Possible effects of climate change on ixodid ticks and the pathogens they transmit: Predictions and observations. J. Med. Entomol. 58, 1536–1545 (2020).Article 

Google Scholar 
23.Zając, Z., Sędzikowska, A., Maślanko, W., Woźniak, A. & Kulisz, J. Occurrence and Abundance of Dermacentor reticulatus in the habitats of the ecological corridor of the Wieprz river, eastern Poland. Insects 12, 96 (2021).PubMed 
PubMed Central 
Article 

Google Scholar 
24.Zając, Z., Bartosik, K. & Woźniak, A. Monitoring Dermacentor reticulatus host-seeking activity in natural conditions. Insects 11, 264 (2020).PubMed Central 
Article 

Google Scholar 
25.Global and European temperature—European Environment Agency. https://www.eea.europa.eu/data-and-maps/indicators/global-and-european-temperature/global-and-european-temperature-assessment-1. Accessed 22 July 2021.26.Średnie i sumy miesięczne. Dane meteorologiczne https://meteomodel.pl/dane/srednie-miesieczne/?imgwid=351220495&par=sndp&max_empty=2. Accessed 22 July 2021.27.Vladimirov, L. N. et al. Quantifying the Northward Spread of Ticks (Ixodida) as climate warms in Northern Russia. Atmosphere 12, 233 (2021).ADS 
Article 

Google Scholar 
28.Mierzejewska, E. J., Alsarraf, M., Behnke, J. M. & Bajer, A. The effect of changes in agricultural practices on the density of Dermacentor reticulatus ticks. Vet. Parasitol. 211, 259–265 (2015).PubMed 
Article 
PubMed Central 

Google Scholar 
29.Zając, Z., Woźniak, A. & Kulisz, J. Infestation of dairy cows by ticks Dermacentor reticulatus (Fabricius, 1794) and Ixodes ricinus (Linnaeus, 1758) in eastern Poland. Ann. Parasitol. 66, 87–96 (2020).PubMed 
PubMed Central 

Google Scholar 
30.Estrada-Peña, A. Climate, niche, ticks, and models: What they are and how we should interpret them. Parasitol. Res. 103, 87–95 (2008).Article 

Google Scholar 
31.Süss, J., Klaus, C., Gerstengarbe, F. W. & Werner, P. C. What makes ticks tick? Climate change, ticks, and tick-borne diseases. J. Travel Med. 15, 39–45 (2008).PubMed 
Article 
PubMed Central 

Google Scholar 
32.Paulauskas, A. et al. New localities of Dermacentor reticulatus ticks in the Baltic countries. Ticks Tick Borne Dis. 6, 630–635 (2015).PubMed 
Article 
PubMed Central 

Google Scholar 
33.Kubiak, K. et al. Dermacentor reticulatus ticks (Acari: Ixodidae) distribution in north-eastern Poland: An endemic area of tick-borne diseases. Exp. Appl. Acarol. 75, 289–298 (2018).CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
34.Silaghi, C., Weis, L. & Pfister, K. Dermacentor reticulatus and Babesia canis in Bavaria (Germany): A georeferenced field study with digital habitat characterization. Pathogens 9, 541 (2020).CAS 
PubMed Central 
Article 

Google Scholar 
35.Kohn, M. et al. Dermacentor reticulatus in Berlin/Brandenburg (Germany): Activity patterns and associated pathogens. Ticks Tick Borne Dis. 10, 191–206 (2019).PubMed 
Article 
PubMed Central 

Google Scholar 
36.Kiewra, D., Czułowska, A., Dyczko, D., Zieliński, R. & Plewa-Tutaj, K. First record of Haemaphysalis concinna (Acari: Ixodidae) in Lower Silesia, SW, Poland. Exp. Appl. Acarol. 77, 449–454 (2019).PubMed 
Article 

Google Scholar 
37.Zieba, P. et al. A new locality of the Haemaphysalis concinna tick (Koch, 1844) in Poland and its role as a potential vector of infectious diseases. Ann. Parasitol. 65, 281–286 (2019).PubMed 

Google Scholar 
38.Gray, J. S., Dautel, H., Estrada-Peña, A., Kahl, O. & Lindgren, E. Effects of climate change on ticks and tick-borne diseases in Europe. Interdiscip. Perspect. Infect. Dis. 2009, 593232 (2009).CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
39.Medlock, J. M. & Leach, S. A. Effect of climate change on vector-borne disease risk in the UK. Lancet Infect. Dis. 15, 721–730 (2015).PubMed 
Article 

Google Scholar 
40.Pfäffle, M., Littwin, N. & Petney, T. Host preferences of immature Dermacentor reticulatus (Acari: Ixodidae) in a forest habitat in Germany. Ticks Tick Borne Dis. 6, 508–515 (2015).PubMed 
Article 
PubMed Central 

Google Scholar 
41.Zając, Z., Bartosik, K., Kulisz, J. & Woźniak, A. Ability of adult Dermacentor reticulatus ticks to overwinter in the temperate climate zone. Biology 9, 145 (2020).PubMed Central 
Article 

Google Scholar 
42.Kiewra, D., Czułowska, A. & Lonc, E. Winter activity of Dermacentor reticulatus (Fabricius, 1794) in the newly emerging population of Lower Silesia, south-west Poland. Ticks Tick Borne Dis. 7, 1124–1127 (2016).PubMed 
Article 
PubMed Central 

Google Scholar 
43.Buczek, A., Bartosik, K. & Zając, Z. Changes in the activity of adult stages of Dermacentor reticulatus (Ixodida: Amblyommidae) induced by weather factors in eastern Poland. Parasites Vectors 7, 245 (2014).PubMed 
PubMed Central 
Article 

Google Scholar 
44.Hubálek, Z., Halouzka, J. & Juricova, Z. Host-seeking activity of ixodid ticks in relation to weather variables. J. Vector Ecol. 28, 159–165 (2003).PubMed 
PubMed Central 

Google Scholar 
45.Bartosik, K., Wiśniowski, Ł & Buczek, A. Questing behavior of Dermacentor reticulatus adults (Acari: Amblyommidae) during diurnal activity periods in eastern Poland. J. Med. Entomol. 49, 859–864 (2012).CAS 
PubMed 
Article 
PubMed Central 

Google Scholar 
46.Buczek, A., Bartosik, K., Wisniowski, L. & Tomasiewicz, K. Changes in population abundance of adult Dermacentor reticulatus (Acari: Amblyommidae) in long-term investigations in eastern Poland. Ann. Agric. Environ. Med. 20, 269–272 (2013).PubMed 
PubMed Central 

Google Scholar 
47.Mierzejewska, E. J., Estrada-Peña, A., Alsarraf, M., Kowalec, M. & Bajer, A. Mapping of Dermacentor reticulatus expansion in Poland in 2012–2014. Ticks Tick Borne Dis. 7, 94–106 (2016).PubMed 
Article 
PubMed Central 

Google Scholar 
48.Zając, Z. et al. Environmental determinants of the occurrence and activity of Ixodes ricinus ticks and the prevalance of tick-borne diseases in eastern Poland. Sci. Rep. 11, 15472 (2021).PubMed 
PubMed Central 
Article 
CAS 

Google Scholar 
49.Kulisz, J., Bartosik, K., Zając, Z., Woźniak, A. & Kolasa, S. Quantitative parameters of the body composition influencing host seeking behavior of Ixodes ricinus adults. Pathogens 10, 706 (2021).PubMed 
PubMed Central 
Article 

Google Scholar 
50.Alasmari, S. & Wall, R. Metabolic rate and resource depletion in the tick Ixodes ricinus in response to temperature. Exp. Appl. Acarol. 83, 81–93 (2020).PubMed 
PubMed Central 
Article 
CAS 

Google Scholar 
51.Zajac, Z., Bartosik, K. & Buczek, A. Factors influencing the distribution and activity of Dermacentor reticulatus (F.) ticks in an anthropopressure-unaffected area in central-eastern Poland. Ann. Agric. Environ Med. 23, 270–275 (2016).PubMed 
Article 

Google Scholar 
52.Bogdaszewska, Z. Range and ecology of Dermacentor reticulatus (Fabricius, 1794) in Mazuria focus. II. Seasonal activity patterns of the adults. Wiad. Parazytol. 50, 731–738 (2004).PubMed 

Google Scholar 
53.Razumova, I. V. The activity of Dermacentor reticulatus Fabr. (Ixodidae) ticks in nature. Med. Parasitol. Parasites Dis. 4, 8–14 (1999).
Google Scholar 
54.Szymański, S. Seasonal activity of Dermacentor reticulatus (Fabricius, 1794) (Acarina, Ixodidae) in Poland I. Adults. Acta Parasitol. Pol. 31, 247–255 (1987).
Google Scholar 
55.Hornok, S. Allochronic seasonal peak activities of Dermacentor and Haemaphysalis spp. under continental climate in Hungary. Vet. Parasitol. 163, 366–369 (2009).PubMed 
Article 
PubMed Central 

Google Scholar 
56.Randolph, S. E. & Storey, K. Impact of microclimate on immature tick-rodent host interactions (Acari: Ixodidae): Implications for parasite transmission. J. Med. Entomol. 36, 741–748 (1999).CAS 
PubMed 
Article 
PubMed Central 

Google Scholar 
57.Nowak-Chmura, M. Ticks (Ixodida) of Central Europe (Pedagogical University of Cracow Press, 2013).
Google Scholar  More

Altogether, we found that, while traffic volume declined by  > 7% during the pandemic year (with a maximum monthly decline of nearly 40%), the absolute number of annual WVCs was largely unchanged. This resulted in significant increases of  > 8% in collision rates between vehicles and wildlife during the pandemic year, peaking at a  > 27% nationwide increase in April 2020. Other studies from the first several months of the pandemic documented similar transient declines in the number of WVCs when the pandemic began which then reversed in many jurisdictions as the pandemic progressed and traffic rebounded26,27. We observed a similar pattern over the first five months of the pandemic at the national scale (Fig. 2): WVCs initially declined during the pandemic in step with declines in traffic volume, but then started to increase to baseline levels at a faster rate than traffic, possibly due to behavioral lags by wildlife following traffic-mediated increases in wildlife road use. Though based on coarse-scale data, our research aligns with assertions from studies during27 and prior to the pandemic3,15,16,28,29 that the relationship between traffic volume and WVCs is non-linear.We postulate that the observed non-linear relationship between traffic volume and WVCs is the result of greater use of roads and roadsides by certain wildlife species, namely large mammals (Table S1), in response to decreasing traffic volume, as prior research has suggested3,14,15,16. This explanation is consistent with accounts of various wildlife species making increased use of human spaces during the pandemic17,20,21: with less cars on the roads, wildlife might be less deterred from roads by the noise and light pollution that accompany high traffic volumes9,10,11,20 and perceive roads as less risky, thereby increasing their willingness to attempt road crossings3,8,15,16. Beyond incidentally crossing roads while moving about the landscape8,9, wildlife might be attracted to roads for travel, mates, or other resources8,10,11. Many animals are shown to utilize roads to move efficiently across the landscape11,12, and roads and the surrounding areas are comparatively open, such that wildlife might select roads and roadsides for enhanced visibility to find mates, detect predators, or locate prey10,13. Roadsides also can provide foraging opportunities and essential nutrients for wildlife via abundant, high-quality early successional vegetation and high salt concentrations10,11. As such, decreased road traffic during the pandemic might have caused certain wildlife species to tolerate the risks associated with roads in order to access the benefits of roads and roadsides.An alternative explanation for the observed increases in collision rates is that human driving behavior, rather than animal behavior, changed during the pandemic. With fewer cars on the road, people might drive faster35, rendering it more difficult for both humans and wildlife to avoid collisions3. Preliminary studies from throughout the United States have indeed suggested changes to human driving behavior during the pandemic, with several jurisdictions reporting increased vehicle speeds35,36. Despite reported increases in vehicle speeds, however, the total number of vehicle collisions (the sum of both wildlife and non-wildlife collisions) mirrored trends in traffic volume and declined considerably during the pandemic37,38. Thus, because changes to human behavior appear to have had a minimal effect on vehicle collisions overall, it is unlikely that the observed changes in collision rates are due to increased vehicle speeds alone. Still, we cannot discount the possibility that changes to human driving behavior contributed to the patterns documented here, and future work should more explicitly test the relative effects of changes in traffic volume on both human driving behavior and wildlife space-use, as well as the resultant impacts on WVCs.A greater understanding of human driving behavior would also help explain our findings regarding changes in traffic patterns during the pandemic. Nationwide, the severity of COVID-19 restrictions accounted for a large amount of the variation in changes in monthly traffic volume (R2 = 0.968), but the severity of restrictions was less influential on changes in yearly traffic across states (Tables S3 and S4). Restrictions implemented throughout the pandemic were largely enacted for the purpose of minimizing travel, and other research has demonstrated that these restrictions were effective at reducing human mobility18,21. Our state-level findings, however, imply that it was not only the restrictions themselves that reduced travel, but possibly also the associated anxiety regarding the risk of contracting the SARS-CoV-2 virus, as has been suggested in other studies21,22,23,24; although we observed the greatest declines in traffic volume early in the pandemic (Fig. 2A) when restrictions were most stringent (Fig. S2)21, there was widespread anxiety about the risks posed by SARS-CoV-2 during this time22,23, which likely motivated people to stay home independent of restrictions24. Indeed, anxiety and risk perception might explain the relationship between traffic volume and the other covariates in our top models (Table S4). Declines in traffic were greatest in the most densely populated states (Fig. 4A) and in states that had the highest and the lowest disease burdens (Fig. 4B). The risk of SARS-CoV-2 transmission is greater in more densely populated states due to the close proximity of and frequent interactions amongst people21. As such, people may have altered their road use more in densely populated states as compared to sparsely populated ones due to differing perceptions of disease transmission risk23—though differences in infrastructure in relation to population density likely contributed to this pattern as well39. Similarly, declines in traffic volume in states with larger outbreaks of SARS-CoV-2 might have been driven by increases in the perceived risk of contracting the virus21,23. Alternatively, traffic reductions in states with low disease burdens might reflect increased compliance with stay-at-home orders, and therefore less opportunity for disease spread40,41; essentially, reductions in traffic volume might be the cause of locally low disease burdens therein, rather than a consequence. Altogether, we posit that the observed heterogeneity in traffic volume between states is, at least in part, attributed to differences in the perceived risk posed by the SARS-CoV-2 virus.Regardless of the mechanisms underlying changes in traffic volume and WVCs, our observation that the annual number of WVCs was largely unchanged despite substantive declines in traffic volume has implications for mitigating WVCs going forward. Most directly, the lack of a directional change in WVCs suggests that road traffic levels in the United States are currently such that even large decreases in traffic volume would have minimal long-term effects on the absolute number of WVCs. As such, decreasing collisions by reducing traffic volume would require even larger and longer-lasting changes in traffic than those observed during the pandemic. Since such massive and sustained reductions in traffic are unlikely4,5,6, WVCs in the United States essentially represent a fixed cost as of now, both for human society and wildlife populations. As such, these transient decreases in traffic likely provided minimal reprieve to large mammals from collision-induced mortality, in contrast to speculation that changes in human mobility during the COVID-19 pandemic had substantial positive effects for wildlife populations by freeing wildlife from the pervasive direct and indirect effects of humans17,18,19,20,26,27,42.Indeed, it is possible that short-term decreases in traffic volume might ultimately be harmful to those wildlife species that increased their road use. Although the increases in collision rates we observed at the beginning of the pandemic were rapid and corresponded to nationwide declines in traffic volume (see also26,27), collision rates remained elevated even as traffic approached baseline levels in July (Fig. 2B). If wildlife responses to changes in traffic are asymmetric (i.e., increases in wildlife road use following declines in traffic occur more rapidly than decreases in wildlife road use in response to increased traffic), then short-term declines in traffic volume might lead to net increases in the number WVCs over longer timeframes, ultimately proving detrimental to certain wildlife populations1,3. Future work should evaluate the long-term effects of the pandemic on wildlife populations, specifically with regards to collision-induced mortality17,20,26,27,42.Although the COVID-19 pandemic provided an opportunity to examine the short-term effects of transient decreases in traffic volume on WVCs, the longer-term effects of expanding human populations, greater road densities, and altogether higher traffic volumes on WVCs are less clear. Similar to the increases in wildlife road use in response to decreases in traffic volume theorized here, steady increases in traffic might reduce wildlife road use long-term3,14,15,16; since road traffic is indeed increasing through time4,5,6, we might therefore see declines in WVCs as roads become more effective at repelling wildlife1,3,14. Although these reductions in vehicle-induced wildlife mortality are welcome, this would see roads increasingly serve as barriers to animal movement and gene flow43, further fragmenting already disconnected wildlife populations8. Thus, policy makers and urban planners should invest in infrastructure such as overpasses, underpasses, and fencing that enables wildlife to cross high-traffic roads safely or directs wildlife towards low-risk areas8,9. Even substantive short-term declines in road traffic are not sufficient to mitigate wildlife-vehicle conflict on their own. More