Ecology

Genetic structure of the parental populationBased on the lnPD and ΔK values obtained using STRUCTURE, we identified two genetic groups within the DHS Acer population (Supplementary Fig. S1). The q value from STRUCTURE analysis represents the proportion of ancestral origin28 (Fig. 2a). Among the 70 individual trees, 72.9% were assigned a q value smaller than 0.1 or larger than 0.9, thereby signifying a typical bimodal distribution (Fig. 2b). Individuals with q value greater than 0.9 and consistent genetic origin from the NEA region were defined as the NEA lineage (hereafter “NEA-DHS”), whereas those with values less than 0.1 and with consistent genetic origin from the SEA region were defined as the SEA lineage (hereafter “SEA-DHS”). Individuals with intermediate q value between 0.1 and 0.9 were defined as hybrid genetic types (hereafter “Hybrid-DHS”). Accordingly, we identified 27 SEA-DHS (38.6%), 24 NEA-DHS (34.3%), and 19 Hybrid-DHS (27.1%) (Fig. 2b).Figure 2Genetic structure of the parental and offspring population. (a) Bar plots illustrating the genetic composition of the adult (leaf) and offspring (fruit) populations in the Daheishan National Nature Reserve (DHS). Each individual is represented by a line partitioned into color segments corresponding to its ancestral proportion. Red color represents the ancestral proportion of Southern East Asia lineage. Green color represents the ancestral proportion of Northern East Asia lineage. Black lines in bar plots of leaf population separate individuals with ancestral proportion (q value) bigger than 0.9 or smaller than 0.1 from hybrids (0.1  0.5) produced by the SEA-DHS were obtained from a single tree, which was identified as SEA-DHS based on the DHS-only dataset, although it was indicated to be Hybrid-DHS based on the whole-range dataset. The Hybrid-DHS maternal trees produced 17.6% pure SEA-DHS seeds, 57.6% pure NEA-DHS seeds, and 24.7% hybrid seeds.Flowering phenologyThe sexual system of Acer has four phenotypes: duodichogamous, protogynous, protandrous, and male31. Hence, there are three functional sex types: (1) “Male I” flowers open earlier than “Female” flowers, with mature stamens, no style, and ovary; (2) “Female” flowers have mature pistils, short filaments, and indehiscence anthers; (3) “Male II” flowers open later than “Female” flowers, with mature stamens, ovaries, and separated stigmas. Duodichogamy is characterized by “Male I,” “Female,” and “Male II” types; protandry by “Male I” and “Female” types; and protogyny by “Female” and “Male II” types31.During the flowering season, we monitored a total of 10,074 flowers produced by 29 trees (Fig. 2d), among which one tree (SEA-DHS) was protandrous, four trees (three Hybrid-DHS and one NEA-DHS) were protogynous, and the remaining 24 trees were duodichogamous. We observed that the blooming phenology of SEA-DHS and NEA-DHS differed significantly to most assessed phenological indices, with a single exception being a marginally significant difference in the peak blooming time of Male I (Table 1). Compared with NEA-DHS, SEA-DHS were characterized by significantly later flowering phenology, with Male I commencement and cessation of blooming being on average two and three days later, respectively. Similarly, the commencement, peak, and cessation of Female occurred later by averages of 4, 4, and 5 days, respectively, whereas those of Male II occurred later by 5, 4, and 5 days, respectively. Furthermore, the duration of blooming was significantly longer in the SEA-DHS group than in the NEA-DHS group by three days. In the case of Hybrid-DHS, the values obtained for all assessed phenological indices were intermediate between those of the two parental types. Among these, the values of the six indices differed significantly from one or the other parental types, with the majority (5/6) differing from those of the SEA-DHS. Thus, phenologically, Hybrid-DHS appeared to be closer to NEA-DHS.Table 1 Flowering phenology of SEA-DHS, Hybrid-DHS, and NEA-DHS.Full size tableHowever, despite the differing phenology of the SEA-DHS and NEA-DHS, we observed instances of overlap in the blooming periods of male or female flowers in one genetic type with those of flowers of the opposite sex in another genetic type. For example, the peak of Female among NEA-DHS (11.67 ± 0.67) was found to coincide with the peak of Male I (11.44 ± 1.06; p = 0.879) in SEA-DHS. Similarly, Female blooming in the SEA-DHS peaked (16.11 ± 1.09) just 1 d after the peak of Male II (15.50 ± 0.43) in the NEA-DHS (p = 0.667), which at this time still retained an abundance of male flowers in bloom. In contrast, we detected no overlapping phenology with respect to the blooming of Male I of NEA-DHS or Male II of SEA-DHS with the Female in another genetic type.Morphological variation of leaves and fruitLeaves Among the eight leaf indices, all except InfectionRatio were significantly different between lineages. Generally, the leaves of NEA-DHS were found to have seven lobes, whereas those of SEA-DHS were typically five lobed (Lobes#), thereby contributing to significantly larger leaves in NEA-DHS than in SEA-DHS (TotalArea). Furthermore, NEA-DHS leaves had shorter and wider central lobes (CentralLength and CentralWidth), as well as an earlier and narrower inflection of the central lobes (InflectionLength and InflectionWidth), compared with those of SEA-DHS (Table 2). Six indices had correlation coefficients of less than 0.7, which were used for principal component analysis (PCA) analysis (Supplementary Table S2). The first two axes of the PCA were found to explain 63.7% of the variation in leaf morphology (Fig. 3a), with InflectionLength, CentralLength, and CentralRatio contributing the most to the first axis (38.2%), whereas TotalArea contributed the most to the second axis (25.5%) (Supplementary Table S3). The leaves of SEA-DHS and NEA-DHS plants were largely clustered in separate groups (Fig. 3a). However, all indices were continuous variables with large overlaps between the lineages (Table 2). For example, NEA-DHS had a significantly larger leaf area (21.06–88.70 cm2) than SEA-DHS (11.34–70.09 cm2). The shape of the central lobe is another major leaf trait that distinguishes between the two species. NEA-DHS had a shorter and wider central lobe (CentralRatio:0.67–2.49), while SEA-DHS had a longer and narrower central lobe (CentralRatio:0.9–3.46).Table 2 Morphological variation in the leaves and fruits of Acer trees in the Daheishan National Nature Reserve.Full size tableFigure 3Morphological variation in the leaves (a) and fruits (b) of southern and northern East Asia lineages of the Acer species complex in the Daheishan National Nature Reserve based on principal component analysis. SEA-DHS: Southern East Asia lineage of the Acer species complex in the DHS; NEA-DHS: Northern East Asia lineage of the Acer species complex in the DHS; Hybrid-DHS: hybrids between SEA-DHS and NEA-DHS lineages.Full size imageWith regard to Hybrid-DHS, the leaves were morphologically intermediate between those of the two parental types (Fig. 3a), as were the values of the assessed morphological trait indices (Table 2).Fruits 11 indices of fruits were significantly different between lineages. NEA-DHS tend to be characterized by smaller fruits (FruitLength and FruitWidth), seeds (SeedLength, SeedWidth and JunctionWidth), and fruit wings (WingLength and WingWidth). Moreover, the seed wings of NEA-DHS fruits are typically oriented at an obtuse angle, whereas those of SEA-DHS fruits tend to be aligned at a right angle (FruitAngle). The length ratio of the wing and seed (Wing:Seed) was larger in NEA-DHS than in SEA-DHS (1.24 vs 1.06, respectively, Table 2). Eight indices had correlation coefficients of less than 0.7, which were retained for PCA analysis (Supplementary Table S4). The first two axes of the PCA explained 58.4% of the variation in fruit morphology (Fig. 3b), with JunctionWidth and SeedLength contributing the most to the first axis (35.1%), whereas SeedRatio and WingRatio contributed the most to the second axis (23.3%) (Supplementary Table S3). The fruits of SEA-DHS and NEA-DHS plants were largely clustered in separate groups, with most fruits of SEA-DHS having negative values in Axis 1, while most fruits of NEA-DHS having positive values (Fig. 3b). Both JunctionWidth and SeedLength in Axis 1 reflect the size of the seed. NEA-DHS had smaller seed (SeedLength: 0.63–1.21 cm, SeedWidth:0.43–0.75 cm), while larger seed in SEA-DHS (SeedLength:0.79–1.49 cm, SeedWidth:0.49–0.93 cm). All indices were continuous variables with large overlaps between the lineages (Table 2).The morphology of Hybrid-DHS fruits was generally intermediate between that of the two parental types (Fig. 3b), as reflected in the values of the different morphological traits. The exceptions in this regard were FruitLength, WingLength, as well as two ratio indices (SeedRatio and WingRatio), with hybrid trees typically producing longer fruit with longer fruit wings (Table 2).Ecological niche divergence between NEA and SEAWe found a positive correlation between q value from Structure analysis and altitude (Pearson’s r = 0.83, p  670 m), whereas SEA-DHS was clustered at the foothill ( More

Ecological networks: Linking structure to dynamics in food webs. (Oxford University Press, 2006).Adaptive food webs: Stability and transitions of real and model ecosystems. (Cambridge University Press, 2018).Pimm, S. L. Food Webs (Springer, 1982).Book 

Google Scholar 
Adaptive Food Webs: Stability and Transitions of Real and Model Ecosystems. (Cambridge University Press, 2017). doi:https://doi.org/10.1017/9781316871867.da Mata, A. S. Complex Networks: A Mini-review. Braz. J. Phys. 50, 658–672 (2020).ADS 
Article 

Google Scholar 
Zhang, W. Fundamentals of Network Biology. (World Scientific (Europe), 2018). https://doi.org/10.1142/q0149.Reichman, O. J., Jones, M. B. & Schildhauer, M. P. Challenges and opportunities of open data in ecology. Science 331, 703–705 (2011).ADS 
CAS 
PubMed 
Article 

Google Scholar 
Farley, S. S., Dawson, A., Goring, S. J. & Williams, J. W. situating ecology as a big-data science: Current advances, challenges, and solutions. Bioscience 68, 563–576 (2018).Article 

Google Scholar 
Osawa, T. Perspectives on biodiversity informatics for ecology. Ecol. Res. 34, 446–456 (2019).Article 

Google Scholar 
Shin, N. et al. Toward more data publication of long-term ecological observations. Ecol. Res. 35, 700–707 (2020).Article 

Google Scholar 
Pringle, R. M. & Hutchinson, M. C. Resolving food-web structure. Annu. Rev. Ecol. Evol. Syst. 51, 55–80 (2020).Article 

Google Scholar 
Derocles, S. A. P. et al. Biomonitoring for the 21st Century: Integrating Next-Generation Sequencing Into Ecological Network Analysis. in Advances in Ecological Research vol. 58 1–62 (Elsevier, 2018).Vacher, C. et al. Learning ecological networks from next-generation sequencing data. in Advances in Ecological Research vol. 54, 1–39 (Elsevier, 2016).Evans, D. M., Kitson, J. J. N., Lunt, D. H., Straw, N. A. & Pocock, M. J. O. Merging DNA metabarcoding and ecological network analysis to understand and build resilient terrestrial ecosystems. Funct. Ecol. 30, 1904–1916 (2016).Article 

Google Scholar 
Pocock, M. J. O. et al. A vision for global biodiversity monitoring with citizen science. in Advances in Ecological Research vol. 59, 169–223 (Elsevier, 2018).Sultana, M. & Storch, I. Suitability of open digital species records for assessing biodiversity patterns in cities: A case study using avian records. J. Urban Ecol. 7, juab014 (2021).Article 

Google Scholar 
Amano, T., Lamming, J. D. L. & Sutherland, W. J. Spatial gaps in global biodiversity information and the role of citizen science. Bioscience 66, 393–400 (2016).Article 

Google Scholar 
Chandler, M. et al. Contribution of citizen science towards international biodiversity monitoring. Biol. Conserv. 213, 280–294 (2017).Article 

Google Scholar 
Fontaine, C. et al. The ecological and evolutionary implications of merging different types of networks: Merging networks with different interaction types. Ecol. Lett. 14, 1170–1181 (2011).PubMed 
Article 

Google Scholar 
Martinson, H. M. & Fagan, W. F. Trophic disruption: A meta-analysis of how habitat fragmentation affects resource consumption in terrestrial arthropod systems. Ecol. Lett. 17, 1178–1189 (2014).PubMed 
Article 

Google Scholar 
Marczak, L. B., Thompson, R. M. & Richardson, J. S. Meta-analysis: Trophic level, Habitat, and productivity shape the food web effects of resource subsidies. Ecology 88, 140–148 (2007).PubMed 
Article 

Google Scholar 
McCary, M. A., Mores, R., Farfan, M. A. & Wise, D. H. Invasive plants have different effects on trophic structure of green and brown food webs in terrestrial ecosystems: A meta-analysis. Ecol. Lett. 19, 328–335 (2016).PubMed 
Article 

Google Scholar 
Cirtwill, A. R., Stouffer, D. B. & Romanuk, T. N. Latitudinal gradients in biotic niche breadth vary across ecosystem types. Proc. R. Soc. B Biol. Sci. 282, 20151589 (2015).Article 
CAS 

Google Scholar 
Fortuna, M. A., Ortega, R. & Bascompte, J. The Web of Life. ArXiv14032575 Q-Bio (2014).Brose, U. et al. Predator traits determine food-web architecture across ecosystems. Nat. Ecol. Evol. 3, 919–927 (2019).PubMed 
Article 

Google Scholar 
Mace, G. M., Norris, K. & Fitter, A. H. Biodiversity and ecosystem services: A multilayered relationship. Trends Ecol. Evol. 27, 19–26 (2012).PubMed 
Article 

Google Scholar 
Keyes, A. A., McLaughlin, J. P., Barner, A. K. & Dee, L. E. An ecological network approach to predict ecosystem service vulnerability to species losses. Nat. Commun. 12, 1586 (2021).ADS 
CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Peng, J. et al. Linking ecosystem services and circuit theory to identify ecological security patterns. Sci. Total Environ. 644, 781–790 (2018).ADS 
CAS 
PubMed 
Article 

Google Scholar 
Su, Y. et al. Modeling the optimal ecological security pattern for guiding the urban constructed land expansions. Urban For. Urban Green. 19, 35–46 (2016).Article 

Google Scholar 
Kowarik, I. Novel urban ecosystems, biodiversity, and conservation. Environ. Pollut. 159, 1974–1983 (2011).CAS 
PubMed 
Article 

Google Scholar 
Di Marco, M., Watson, J. E. M., Venter, O. & Possingham, H. P. Global biodiversity targets require both sufficiency and efficiency. Conserv. Lett. 9, 395–397 (2016).Article 

Google Scholar 
Kim, K.-H. & Pauleit, S. Landscape character, biodiversity and land use planning: The case of Kwangju City Region, South Korea. Land Use Policy 24, 264–274 (2007).Article 

Google Scholar 
Young, J. et al. Towards sustainable land use: Identifying and managing the conflicts between human activities and biodiversity conservation in Europe. Biodivers. Conserv. 14, 1641–1661 (2005).Article 

Google Scholar 
Dardonville, M., Urruty, N., Bockstaller, C. & Therond, O. Influence of diversity and intensification level on vulnerability, resilience and robustness of agricultural systems. Agric. Syst. 184, 102913 (2020).Article 

Google Scholar 
Oliver, T. H. et al. Biodiversity and resilience of ecosystem functions. Trends Ecol. Evol. 30, 673–684 (2015).PubMed 
Article 

Google Scholar 
Lau, M. K., Borrett, S. R., Baiser, B., Gotelli, N. J. & Ellison, A. M. Ecological network metrics: Opportunities for synthesis. Ecosphere 8, e01900 (2017).Article 

Google Scholar 
Newman, M. E. J. Networks. (Oxford University Press, 2018).Levine, S. Several measures of trophic structure applicable to complex food webs. J. Theor. Biol. 83, 195–207 (1980).ADS 
Article 

Google Scholar 
Guimarães, P. R. The structure of ecological networks across levels of organization. Annu. Rev. Ecol. Evol. Syst. 51, 433–460 (2020).Article 

Google Scholar 
Dormann, C. F., Frund, J., Bluthgen, N. & Gruber, B. Indices, graphs and null models: Analyzing bipartite ecological networks. Open Ecol. J. 2, 7–24 (2009).Article 

Google Scholar 
Jordán, F., Benedek, Z. & Podani, J. Quantifying positional importance in food webs: A comparison of centrality indices. Ecol. Model. 205, 270–275 (2007).Article 

Google Scholar 
Jordán, F., Liu, W. & Davis, A. J. Topological keystone species: Measures of positional importance in food webs. Oikos 112, 535–546 (2006).Article 

Google Scholar 
Jordán, F., Okey, T. A., Bauer, B. & Libralato, S. Identifying important species: Linking structure and function in ecological networks. Ecol. Model. 216, 75–80 (2008).Article 

Google Scholar 
Jiang, L. Determination of keystone species in CSM food web: A topological analysis of network structure. Netw. Biol. 5, 13 (2015).
Google Scholar 
Abarca-Arenas, L. G., Franco-Lopez, J., Peterson, M. S., Brown-Peterson, N. J. & Valero-Pacheco, E. Sociometric analysis of the role of penaeids in the continental shelf food web off Veracruz. Mexico Based By-catch Fish. Res. 87, 46–57 (2007).
Google Scholar 
Abascal-Monroy, I. M. et al. Functional and structural food web comparison of Terminos Lagoon, Mexico in Three Periods (1980, 1998, and 2011). Estuaries Coasts 39, 1282–1293 (2016).Article 

Google Scholar 
McDonald-Madden, E. et al. Using food-web theory to conserve ecosystems. Nat. Commun. 7, 10245 (2016).ADS 
CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Windsor, F. M. et al. Identifying plant mixes for multiple ecosystem service provision in agricultural systems using ecological networks. J. Appl. Ecol. 58, 2770–2782 (2021).Article 

Google Scholar 
Klaise, J. & Johnson, S. The origin of motif families in food webs. Sci. Rep. 7, 16197 (2017).ADS 
PubMed 
PubMed Central 
Article 
CAS 

Google Scholar 
Estrada, E. Characterization of topological keystone species. Ecol. Complex. 4, 48–57 (2007).Article 

Google Scholar 
Thompson, R. M. & Townsend, C. R. Impacts on stream food webs of native and exotic forest: An intercontinental comparison. Ecology 84, 145–161 (2003).Article 

Google Scholar 
Bascompte, J., Melian, C. J. & Sala, E. Interaction strength combinations and the overfishing of a marine food web. Proc. Natl. Acad. Sci. 102, 5443–5447 (2005).ADS 
CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Dunne, J. A. et al. The roles and impacts of human hunter-gatherers in North Pacific marine food webs. Sci. Rep. 6, 21179 (2016).ADS 
CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Gauzens, B., Legendre, S., Lazzaro, X. & Lacroix, G. Food-web aggregation, methodological and functional issues. Oikos 122, 1606–1615 (2013).Article 

Google Scholar 
Patonai, K. & Jordán, F. Aggregation of incomplete food web data may help to suggest sampling strategies. Ecol. Model. 352, 77–89 (2017).Article 

Google Scholar 
Thompson, R. M. & Townsend, C. R. Is resolution the solution?: The effect of taxonomic resolution on the calculated properties of three stream food webs. Freshw. Biol. 44, 413–422 (2000).Article 

Google Scholar 
Abarca-Arenas, L. G. & Ulanowicz, R. E. The effects of taxonomic aggregation on network analysis. Ecol. Model. 149, 285–296 (2002).Article 

Google Scholar 
Jordán, F. & Osváth, G. The sensitivity of food web topology to temporal data aggregation. Ecol. Model. 220, 3141–3146 (2009).Article 

Google Scholar 
European Commission. Communication from the commission to the european parliament, the council, the european economic and social committee and the committee of the regions: EU Biodiversity Strategy for 2030 Bringing nature back into our lives. Preprint at https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:52020DC0380 (2020).European Parliament. European Parliament resolution of 9 June 2021 on the EU Biodiversity Strategy for 2030: Bringing nature back into our lives (P9_TA(2021)0277). Preprint at https://www.europarl.europa.eu/doceo/document/TA-9-2021-0277_EN.html (2021).Felson, A. J. & Ellison, A. M. Designing (for) Urban Food Webs. Front. Ecol. Evol. 9, 582041 (2021).Article 

Google Scholar 
Warren, P. et al. Urban food webs: Predators, prey, and the people who feed them. Bull. Ecol. Soc. Am. 87, 387–393 (2006).Article 

Google Scholar 
De Montis, A., Ganciu, A., Cabras, M., Bardi, A. & Mulas, M. Comparative ecological network analysis: An application to Italy. Land Use Policy 81, 714–724 (2019).Article 

Google Scholar 
Poisot, T. et al. Mangal—making ecological network analysis simple. Ecography 39, 384–390 (2016).Article 

Google Scholar 
Morris, Z. B., Weissburg, M. & Bras, B. Ecological network analysis of urban–industrial ecosystems. J. Ind. Ecol. 25, 193–204 (2021).Article 

Google Scholar 
Chamberlain, S. A. & Szöcs, E. taxize: Taxonomic search and retrieval in R. F1000 Research 2, 191 (2013).PubMed 
PubMed Central 
Article 

Google Scholar 
Hagberg, A. A., Schult, D. A. & Swart, P. J. Exploring network structure, dynamics, and function using networkX. in Proceedings of the 7th Python in Science Conference (eds. Varoquaux, G., Vaught, T. & Millman, J.) 11–15 (2008).Scotti, M. & Jordán, F. Relationships between centrality indices and trophic levels in food webs. Community Ecol. 11, 59–67 (2010).Article 

Google Scholar 
Gouveia, C., Móréh, Á. & Jordán, F. Combining centrality indices: Maximizing the predictability of keystone species in food webs. Ecol. Indic. 126, 107617 (2021).Article 

Google Scholar 
Allesina, S. & Pascual, M. Googling Food Webs: Can an Eigenvector Measure Species’ Importance for Coextinctions?. PLoS Comput. Biol. 5, e1000494 (2009).ADS 
MathSciNet 
PubMed 
PubMed Central 
Article 
CAS 

Google Scholar 
Patro, S. G. K. & Sahu, K. K. Normalization: A preprocessing stage. https://doi.org/10.48550/ARXIV.1503.06462(2015).Reback, J. et al. pandas-dev/pandas: Pandas 1.2.3. (Zenodo, 2021). 10.5281/ZENODO.4572994.Hunter, J. D. Matplotlib: A 2D graphics environment. Comput. Sci. Eng. 9, 90–95 (2007).Article 

Google Scholar 
Waskom, M. et al. mwaskom/seaborn: v0.11.1 (December 2020). (Zenodo, 2020). 10.5281/ZENODO.4379347.Girvan, M. & Newman, M. E. J. Community structure in social and biological networks. Proc. Natl. Acad. Sci. 99, 7821–7826 (2002).ADS 
MathSciNet 
CAS 
PubMed 
PubMed Central 
MATH 
Article 

Google Scholar 
Rosvall, M., Axelsson, D. & Bergstrom, C. T. The map equation. Eur. Phys. J. Spec. Top. 178, 13–23 (2009).Article 

Google Scholar 
Gao, P. & Kupfer, J. A. Uncovering food web structure using a novel trophic similarity measure. Ecol. Inform. 30, 110–118 (2015).Article 

Google Scholar 
Gauzens, B., Thébault, E., Lacroix, G. & Legendre, S. Trophic groups and modules: Two levels of group detection in food webs. J. R. Soc. Interface 12, 20141176 (2015).PubMed 
PubMed Central 
Article 

Google Scholar 
Rudiger, P. et al. holoviz/holoviews: Version 1.14.2. (Zenodo, 2021). 10.5281/ZENODO.4581995.Pedregosa, F. et al. Scikit-learn: Machine learning in Python. J. Mach. Learn. Res. 12, 2825–2830 (2011).MathSciNet 
MATH 

Google Scholar  More

DataSpatial and temporal rangeWe focused on a geographic area that is defined by a cropping window with the corner points P1 (Lon = −180, Lat = 25) and P2 (Lon = −52, Lat = 80), covering the majority of the North American continent (e.g., Fig. 3). We focused on the last 30 Myr, a time span encompassing most of our available sites with paleovegetation information (Supplementary Fig. 1). From the following data sources, we only selected those data points that fall within this spatiotemporal range.Our approach described below required discretizing the input data of past vegetation labels and fossil occurrences into time-bins. For this, we chose the age boundaries of geological stages defined in the International Chronostratigraphic Chart, v2020/0345, since these stages are expected to represent meaningful temporal units for analyzing both faunal and floral patterns. A total of 17 geological stages fell within our selected time frame of the last 30 Myr. We discretized the ages of all data points (vegetation data and fossil occurrences) that fell within a given stage by setting them to the midpoint of the respective stage.Paleovegetation dataWe reviewed a large body of peer-reviewed literature containing paleovegetation reconstructions and compiled a database of 331 sites with paleovegetation data for North America (Supplementary Data 1). These sites represent individual vegetation reconstructions based on fossil evidence (phytoliths, pollen, macrofossil assemblages) of distinct locations in time and space. We condensed the vegetation interpretation of the compiled vegetation data, which in many cases described specific vegetation ecosystem components, into the broader labels “open” versus “closed” vegetation. This resulted in 180 sites being labeled as closed and 151 as open, their dating rounded to the midpoint of the nearest geological stage (Supplementary Data 1). For several of these sites we found multiple vegetation reconstructions in the reviewed literature, for example when multiple sediment samples were taken from the same horizon of a given formation, belonging to the same geological stage. We treated these spatiotemporal duplicates as a single data point, excluding sites with mixed vegetation information (i.e., containing both open and closed vegetation reconstructions).Current vegetation dataTo supplement the limited number of paleovegetation sites, we compiled data about the current vegetation within our study area. In order to obtain current vegetation patterns, we downloaded the SYNMAP Global Potential Vegetation data29. As for the paleovegetation data, we collapsed the more detailed biome data into broader categories by coding the SYNMAP biome IDs  More

China economy-wide material flow identification: system boundary, processes, and materialsThe first step is to define an economy, i.e., the economic (rather than geographical) territory of a country in which the activities and transactions of producer and consumer units are resident. Additionally, the period is a total of thirty-one years, from 1990 to 2020, for the following reasons: (1) statistics before 1990 are of poor quality and are insufficient to allow us to conduct analyses; and (2) so far, statistics have just recently been updated to cover the year of 2020. Furthermore, the analytical framework (hereinafter referred to as China EW-MFA) is developed to explore material utilisation and its environmental consequences within China’s economy.The general structure of China EW-MFA is depicted in Fig. 1, which comprises seven processes. (1) Input of extracted resources: domestic natural resources are extracted from the environment to the economy through human-controlled means. (2) Output of domestic processed materials: after being processed by manufacturers, materials are released from the economy into the environment in the form of by-products and residues, which can be classified by their destinations (i.e., air, land, and water) and pathways (dissipative use and losses). (3) Input and (4) output by cross-border trade: by imports and exports, materials are transported between China’s economy and the economies of the rest of the world. (5) Input and (6) output of balancing items (BI): sometimes, materials identified in the output processes are not considered by inputs, which needs to be balanced. For example, the utilisation of fossil energy materials by combustion causes the emission of carbon dioxide (CO2) into the air, which is identified as system output, but requirements of oxygen (O2) as system input are not counted. (7) Additions to the system: within the economy, materials would have been added to the economy in the form of buildings, infrastructures, durable goods, and household appliances, which are referred to as the net additions to stock (NAS).Fig. 1The general structure of China EW-MFA. To note, white data cells can be obtained directly from official statistics, whereas grey cells are estimated.Full size imageThe last step is to specify the materials concerned in each process. Four types (in blue boxes in Fig. 1) of natural materials are extracted and input into the economy in China, i.e., harvested biomass (33 items), mined metal ores (28 items), quarried non-metallic minerals (155 items), and mined fossil energy materials (6 items in 3 classes). Materials (green boxes) released into the air are greenhouse gases (e.g., CO2, methane (CH4), dinitrogen oxide (N2O)), air pollutants (e.g., particulate matter 10 (PM10), black carbon (BC)), and toxic contaminants of mercury (Hg) in divalent, gaseous elemental, and particulate forms. Those released into the water are inorganic matters (of nitrogen (N), phosphorus (P), Arsenic (As), and four heavy metals of lead (Pb), mercury (Hg), cadmium (Cd), and chromium (Cr)) and organic matters of cyanide, petroleum, and volatile phenol. Materials released into the land are waste disposal in uncontrolled landfills, which are illegal in China. Some materials are dissipated by application, for example, fertilisers, compost, sewage sludge being applied to agricultural land, and pesticides being used to cultivate crops. Some would be unintentionally dissipated from abrasion, corrosion, erosion, and leakages. Materials (in red boxes) are BI, which includes the input of O2 and output of water vapour in the fossil energy material combustion process, the input of O2 and output of water vapour and CO2 in the respiration process of human and cultivated livestock, input and output of water in imported and exported beverages, and the output of water from domestically extracting crops.There are some messages needed to be mentioned: (1) Material of water is not included since its flow volume is more substantial than others, which needs to be independently analysed; (2) Activities of foreign tourists, cross-border transfer of emissions through natural media, etc. are excluded. (3) To be clear, we refer to a data cell as a specific flow process of a specific substance in a specific year, e.g., the number of cereals domestically extracted in 2020.Data acquisition: sources and collectionBased on our China EW-MFA, we first analyse accessibility, reliability, completeness, rules of redistribution, etc., for each data source (yellow boxes in Fig. 1), including China national database, China rural statistical yearbooks, USGS mineral yearbooks, etc. The complete list of data sources and descriptions are presented in Table 1. Then, we store the originally retrieved data source files in a semi- or unstructured format (e.g., CSV, PDF). Next, we manually collect these statistics and reorganise them according to China EW-MFA material types and processes. However, only a tiny part of retrieved statistics can be applied directly, as specified in black colour in Fig. 1.Table 1 Data sources and descriptions.Full size tableData compilation: parameter localisation and data estimationA few inconsistencies in statistics were noticed, which would result in data incompleteness. For example, the domestic extraction of vegetables has been accounted for and published since 1995, before which statistics are unavailable. The domestically harvested timber has been measured in the volume unit of cubic metres, which needs to be converted into the mass unit via density conversion factor. Therefore, acquired statistics have to be estimated, which are specified in grey colour in Fig. 1. The following section elaborates on each data cell’s estimation methods, localised parameters, references, etc. In our uploaded data files, the original statistics, data sources, and compilation methods (using formulas) are all implemented, as explained in the Data Records Section.

The input of natural resources by domestic extraction

Vegetables in crops: Statistics of vegetable production (WVegetables)16 during 1990–1994 are unavailable, which is estimated based on the relationship between the production yield (PYield) and areas (AVegetables), as shown in Eq. 1. Here, PYield is assumed to remain constant at 27.04 thousand tonnes per thousand hectares from 1990 to 1995, derived by dividing vegetable production (257,267 thousand tonnes) by areas (9,515 thousand hectares) in 1995.$${W}_{Vegetables}={P}_{Yield}times {A}_{Vegetables}$$
(1)

Nuts in crops: One of them is chestnuts. The chestnut production in 2020 is unavailable, which is assumed to be the same as in 2019.

Crop residues in biomass residues: They are referred to as that harvested production of crops that do not reach the market to be sold but are instead employed as raw materials for commercial purposes such as energy generation and livestock husbandry. This number (Wcrop residues) can be calculated by first determining the number of crop residues available from primary crop production (Wcrop) and the harvest factor (Pharvest factor), and then using the recovery rate (Precovery rate) to determine the number of crop residues used by the economy, as shown in Eq. 2. These parameters have been localized by previous studies17,18, which are adopted in this study, i.e., wheat (1.1 for Pharvest factor and 0.463 for Precovery rate), maize (1.2, 0.463), rice (0.9, 0.463), sugar cane (0.5, 0.9), beetroots (0.7, 0.9), tuber (0.5, 0.463), pulse (1.2, 0.7), cotton (3.4, 0.463), fibre crops (1.8, 0.463), silkworm cocoons (1.8, 0.463), and oil-bearing crops (1.8, 0.463).$${W}_{cropresidues}={W}_{crop}times {P}_{harvestfactor}times {P}_{recoveryrate}$$
(2)

Roughage of grazed biomass and fodder crops in biomass residues: In China, the grazed biomass for roughage includes annual forage and perennial forage, whereas fodder crops comprise straw feed, processed straw feed, and all other fodder crops. However, information19 on grazed biomass production is only accessible from 2006 to 2018, whereas fodder crop statistics are only available from 2015 to 2017. Equation 3 and Eq. 4 can be used to estimate unavailable statistics. To note, we assume that China’s domestic roughage supply structure has remained unaltered, which has two meanings. The proportion of total domestic roughage production (WDomestic production) in requirement (WRoughage requirement) has remained constant, while the proportion (PSupply fraction) of grazed biomass and fodder crop in domestic roughage production has been unchanged. The requirement (WRoughage requirement) is determined by the quantity of livestock (QLivestock) and their annual feeding amount (PAnnual intake). PAnnual intake (in tonnes per head per year) has been localised for each type of livestock4, with 4.5 for live cattle and buffaloes, 0.5 for sheep and goats, 3.7 for horses, and 2.2 for mules and asses.$${W}_{Roughagerequirement}={Q}_{Livestock}times {P}_{Annualintake}$$
(3)
$${W}_{Domesticproduction}={W}_{Roughagerequirement}times {P}_{Supplyfraction}$$
(4)

Timber in wood: As illustrated in Eq. 5, wood production16 is reported in volume units of cubic metres (VTimber), which need to be converted into mass units (WTimber) via density (PDensity). The parameter PDensity is assumed to be 0.58 tonnes per cubic metre, calculated by averaging 0.52 for coniferous types and 0.64 for non-coniferous ones4.$${W}_{Timber}={V}_{Timber}times {P}_{Density}$$
(5)

Non-ferrous metals in metal ores: Non-ferrous metal statistics are derived from two sources. China statistics20 are measured in gross ore (WMetal ores in gross ore) but are only available from 1999 to 2017, whereas the USGS statistics21 cover the period of 1990 to 2020 but they are measured in metal or concentrate content (WMetal ores in other units). Therefore, USGS statistics need to be converted with an empirical unit conversion factor (PUnit conversion factor) before being applied to estimate unavailable statistics reported by China, as shown in Eq. 6. Conversion factors are localised for each non-ferrous metal in each year from 2000 to 2017 by using USGS statistics divided by China statistics and then averaged after removing the highest value and the lowest value (i.e., trimmed mean). This factor could capture the general relationship between statistics from two separate sources, which can be used in other long time-series studies on resource management on a particular element in China.$${W}_{Metaloresingrossore}={W}_{Metaloresinotherunits}/{P}_{Unitconversionfactor}$$
(6)

Non-metallic minerals: The official China-specific information on non-metallic mineral domestic production is available between 1999 and 201720, the rest of which could be estimated from USGS statistics (1990–2020)21. Also, two differences in reporting standards are observed resulting from the material coverages and reporting units. China statistics contain eighty-eight materials in mineral ores, whereas the USGS only includes twenty in the concentrate unit. Therefore, a conversion factor is developed in this estimation, as shown in Eq. 7. This conversion factor is applied to the total amount of non-metallic mineral production, which is assumed to have been constant from 1990 to 1999 at 11.38% (1999) and 12.56% (2017) from 2017 to 2020.$${W}_{Mineralsingrossore}={W}_{Mineralsinotherunits}/{P}_{Conversionfactor}$$
(7)

Coal in fossil energy materials: Coal, mined in China, includes raw coal, peat, stone coal, and oil shale. Except for raw coal, statistics for the rest are only available from 1999 to 201720. The unavailable data (WOther coals) is estimated using Eq. 8 under the assumption that the structure of the coal supply in China barely changes. That is, the proportion (PSupply fraction) of peat, stone coal, and oil shale in raw coal production (WRaw coal) remains constant, so the 1999 proportion is applied to all years before that (earlier years of 1990–1998), while the 2017 proportion is used to the recent years between 2018 and 2020. For example, PSupply fraction for oil shale production was assumed to be 0.014% during 1990–1999, calculated by dividing raw coal production (1,250,000) by oil shale production (179) in 1999. PSupply fraction in the earlier and the recent years are 0.007% and 0.001% for peat, 0.203% and 0.031% for stone coal, and 0.014% and 0.067% for oil shale.

$${W}_{Othercoals}={W}_{Rawcoal}/{P}_{Supplyfraction}$$
(8)

The output of processed materials by release

Materials released into the air: In China, thirteen materials are released into the air, as shown in Fig. 1. The emission of sulphur dioxide (SO2) is reported in China environmental statistical yearbooks22,23, while the rest is specified in the EDGAR24. However, in EDGAR, statistics for recent years have not yet been updated, which are estimated with the value in the most recent year in our database. For example, nitrous oxide (NOx) records are only available for the years prior to 2016, with 26,365 thousand tonnes in 2015 and 26,837 in 2014. As a result of the observed decreasing trend in NOx emissions, NOx emission data for 2016–2020 is estimated to be 26,000 thousand tonnes. This estimate may be subjective due to constraints, but it would be aligned with European statistics, allowing for international comparisons. Data can be updated after the EDGAR statistics have been updated.

Materials released into the water: Ten principal materials have been found in China wastewater (both industrial and municipal) that are nitrogen (N), phosphorus (P), organic pollutants of petroleum, volatile phenol and cyanide, heavy metals of mercury (Hg), lead (Pb), cadmium (C·d), and the hexavalent chromium (Cr6+), and arsenic (As). Many statistics22,23 have been of poor quality (e.g., inconsistent material coverages between years). Given that the statistics of pollutants in industrial wastewater cover more periods and contain fewer abnormal observations, the total material emissions can be approximated from those of industrial wastewater. Equations 9 and 10 show the estimation processes. The materials in industrial wastewater (WIndustrial materials) are first identified using material mass concentration (PConcentration) and the weight of industrial wastewater (WIndustrial wastewater), and then the materials in total wastewater (WTotal materials) are identified using the proportion (PContribution) of materials in industrial wastewaters (WIndustrial materials) to the total. The assumption is that PConcentration and PContribution change gradually between years, which enables to use linear interpolation method to estimate unavailable parameters. Consider cyanide: its PConcentration was 23.61 (1‰ ppm) in 2005 and 37.31 in 2002, which was assumed to be 28.18 in 2004 and 32.74 in 2003. PConcentration was assumed to be 100% throughout the years for cyanide because all cyanide emissions in China are driven by industrial wastewater discharges. Later, the total material emissions can be derived by dividing the industrial wastewater mass by PConcentration.$${W}_{Industrialmaterials}={W}_{Industrialwastewater}times {P}_{Concentration}$$
(9)
$${W}_{Totalmaterials}={W}_{Industrialmaterials},/,{P}_{Contribution}$$
(10)

Materials released to the land: This is zero because uncontrolled landfills are illegal in China.

Materials dissipated by organic fertiliser use: In China, manure is the primary organic fertiliser, which is excreted by pigs, dairy cows, calves, sheep, horses, asses, mules, camels, chickens, and other animals. As shown in Eq. 11, the manure production (WManure) is estimated through the amounts of raised livestock (QLivestock, heads), the weight of daily manure production (PManure production, kilograms per head per day), the number of days they are raised (PFeeding period, in days per year), and the moisture content of their manure (PDry matter, %) for each type of animal. These parameters are region-specific, which have been localised by Chinese scholars25,26,27 and listed in Table 2.$${W}_{Manure}={Q}_{Livestock}times {P}_{Manureproduction}times {P}_{Feedingperiod}times {P}_{Drymatter}$$
(11)
Table 2 Localised parameters for animal manure production.Full size table

Materials dissipated by mineral fertiliser use: The mineral fertilisers used in China are four types, i.e., nitrogen (N), phosphorus (P), potash (K), and compound. Their usage (WFertiliser usage) is measured in nutrient mass (WNutrient materials), which needs to be converted into the gross mass by dividing their nutrient content (PNutrient content). Equation 12 shows the estimation. This parameter of PNutrient content is localised by the Ministry of Agriculture and Rural Affairs of China28 as 29%, 22%, 35%, and 44% for N- bearing, P- bearing, K-bearing, and compound fertilisers, respectively.$${W}_{Fertiliserusage}={W}_{Nutrientmaterials}/{P}_{Nutrientcontent}$$
(12)

Materials dissipated by sewage sludge: Sewage sludge is the residue generated by municipal wastewater treatment. As demonstrated in Eq. 13, its dissipative use (Wss, dissipation) is the untreated amount of production (Wss, production), represented by the parameter of Pss, dissipation rate. Sewage sludge production (Wss, production) statistics are only available for the years 2006–202029, and data for the remaining years can be estimated using Eq. 14 and Eq. 15. In Eq. 14, Pss, production rate represents the relationship between sewage sludge production (Wss, production, 2006–2020) and wastewater treatment (Www, treatment, 2002–2020), and in Eq. 15, Pww, treatment efficiency represents the relationship between the quantity of treated wastewater (Www, treatment, 2002–2020) and the treatment capacity (Www, treatment capacity, 1990–2020). In this estimation, three assumptions are made. The first is to estimate Www, treatment, Pww, treatment efficiency is assumed to be unchanged at 63% during 1990–2001, given it has been increasing from 63% in 2002 to ~80% in recent years. The second is that, in order to estimate Wss, production, Pss, production rate is assumed to be unchanged at 3.5 between 1990 and 2005, suggesting 3.5 tonnes of sewage sludge are generated by processing 10,000 cubic metres of wastewater. This assumption is determined by that Pss, production rate is approximately 3.5 during 2006–2010 while declines sharply and stabilises at around two during 2011–2020. The last is, to estimate the Wss,dissipation, Pss,dissipation rate is assumed to be 5% between 1990 and 2005, given it has been around 5% during 2006–2020.$${W}_{ss,dissipation}={W}_{ss,production}times {P}_{ss,dissipationrate}$$
(13)
$${W}_{ss,production}={W}_{ww,treatment}times {P}_{ss,productionrate}$$
(14)
$${W}_{ww,treatment}={W}_{ww,treatmentcapacity}times {P}_{ww,treatmentefficiency}$$
(15)

Materials dissipated by composting: Composting is a natural process that uses microbes to turn organic materials into other products, which are then used for fertilising and entering the environment. In China, composting has been used to treat two materials: feces and municipal waste, whose quantities (WComposting) were only available from 2003 to 201029. The unavailable data can be estimated using Eq. 16. The dry weight of materials treated by composting (WComposting) is proportionally related to the fresh weight of all treated materials (WTotal), the proportion treated by composting (PComposting rate), and the dry content (PDry matter). Considering that China’s composting capacity has been declining since 2001 due to the implementation of waste incineration power generation technologies30, Pcomposting rate is assumed to be the same as it was in 2003 (9.5%) between 1990 and 2002, and 1.5% in 2010 between 2011 and 2020. The parameter of PDry matter is 50%4.$${W}_{Composting}={W}_{Total}times {P}_{Compostingrate}times {P}_{Drymatter}$$
(16)

The input and output by cross-border trade. Statistics of imports and exports have been gathered since 1962 and stored in the UN Comtrade database31. However, the data quality issue of outliers, and missing values, especially in weight, is reportedly identified. In our previous work, we addressed these issues, and an improved database32 is provided. Details about our estimation methods can be found in publications33,34,35. As UN Comtrade lists 5,039 different commodity types (in 6-digit HS0 commodity code), yet only 18 material types are specified in the China EW-MFA, UN Comtrade statistics need to be aligned to the China EW-MFA framework. Therefore, we compared each commodity and each material type between them and established a correspondence table to map UN Comtrade commodity types onto our EW-MFA material types. For example, non-ferrous metal materials of China EW-MFA include commodities, such as copper ores and concentrates (260300 HS0 code), silver powder (710610), manganese, articles thereof, and waste or scrap (811100), etc., whereas biomass residues include cereal straw and husks (121300), lucerne meal and pellets (121410), and other fodder and forage products (121410). This correspondence table between HS0 and EW-MFA classification for imports and exports is provided in Supplementary File 1.

The input of balancing items

O2 required for combustion: In BI, requirements for materials can be abstracted as equalling exogenous demands minus intrinsic supplies (Eq. 17). Three parts (two demands and one supply) are considered for O2 requirements by the combustion process: (1) demanding exogenous oxygen to oxidise elements (e.g., carbon, sulphur, nitrogen, etc., except for hydrogen) released into the air, (2) demanding exogenous oxygen to oxidise the hydrogen embedded in fossil energy materials, and (3) providing intrinsic oxygen embedded in fossil energy materials. The first part can be estimated via Eq. 18 by multiplying air emissions (WEmissions) of CO2, N2O, NOx, CO, and SO2 by their oxygen content (POxygen content). For the second (Eq. 19), the oxygen demand is estimated based on the principle of mass balance by converting the hydrogen amount of domestically utilised fossil energy materials (WFossil fuel materials × PHydrogen content) via molar mass conversion factor (PMass conversion factor). PMass conversion factor equals 7.92, derived by the molar mass of one oxygen (16 g/mol) divided by that of two hydrogen atoms (2 × 1.01 g/mol). The last is the intrinsic supplies from fossil fuel materials, which is identified via Eq. 20 by multiplying the domestically utilised amount of fossil fuel materials (WFossil fuel materials) by their oxygen content (POxygen content). The parameters in this estimation are presented in Table 3. As a footnote here, the domestically utilised amount is referred to as the domestic material consumption (DMC), which equals domestic extraction (DE) plus imports (IM) and minus exports (EX).$${W}_{Requirements}={W}_{Demands}-{W}_{Supplies}$$
(17)
$${W}_{Demands}={W}_{Emissions}times {P}_{Oxygencontent}$$
(18)
$${W}_{Demands}={W}_{Fossilfuelmaterials}times {P}_{Hydrogencontent}times {P}_{Massconversionfactor}$$
(19)
$${W}_{Supplies}={W}_{Fossilfuelmaterials}times {P}_{Oxygencontent}$$
(20)
Table 3 Parameters related to combustion processes4.Full size table

O2 required for respiration: O2 is required by the metabolic activities of living organisms, the majority of which are humans and livestock. Bacteria are another sort of organism, which are not included in this estimation because their O2 requirements are too small to be quantified. The respiration-required O2 is related to the total quantity (QOrganisms) and their respiration activity by organism types, as shown in Eq. 21. The respiration activity is represented by the respiration requirement coefficient (PRespiration requirement coefficient), which is the average quantity of O2 that each organism utilises to maintain the metabolic activity, as listed in Table 4.$${W}_{Demands}={Q}_{Organisms}times {P}_{Respirationrequirementcoefficient}$$
(21)
Table 4 Parameters related to respiration processes4.Full size table

Water required for the domestic production of exported beverages: The exported beverages are produced domestically using domestically extracted materials, especially a large amount of water. The weight of water is considered in the output by cross-border trade but is not included in the domestic extraction input. The resulted imbalance can be identified by specifying the water weight in beverages, i.e., multiplying the traded beverage weight (WMaterials) by a parameter of the water content (PWater content), as given in Eq. 22. Fruit and vegetable juices (2009 in HS0 code) and beverages (code 22) are covered in the improved UN Comtrade database32, with PWater content of 85% for the first and 90% for the latter4.

$${W}_{Water}={W}_{Materials}times {P}_{Watercontent}$$
(22)

The output of balancing items.

Water vapour from combustion: Water vapour emissions by domestically combusting fossil fuel materials are contributed by two paths. The direct evaporation of embedded water is the first path (Eq. 23), which can be derived by multiplying the DMC of fossil fuel materials by their moisture content (PMoisture content). The PMoisture content for each type of fossil fuel material is listed in Table 3. The other is the generation of water vapour during hydrogen oxidation, which can be calculated by converting the oxidised weight of hydrogen to the water weight using the molar mass conversion factor (PMass conversion factor), as given in Eq. 24. PMass conversion factor equals 8.92 by dividing the molar mass of water (18.02 g/mol) by that of two hydrogen atoms (2 × 1.01 g/mol).$${W}_{Water}={W}_{Fossilfuelmaterials}times {P}_{Moisturecontent}$$
(23)
$${W}_{Water}={W}_{Fossilfuelmaterials}times {P}_{Hydrogencontent}times {P}_{Massconversionfactor}$$
(24)

Water vapour and CO2 from respiration: Respiration activities of organisms will produce water vapour and CO2, whose estimation is similar to that of O2 requirements. As shown in Eq. 25, the respiration-caused gas emissions are related to the number of organisms (QOrganisms) and the respiration activity by organism types. The latter is represented by the parameter of respiration emission coefficient (PRespiration emission coefficient), which is specified in Table 4 for water vapour and CO2 for each type of organism.$${W}_{Emissions}={Q}_{Organisms}times {P}_{Respirationemissioncoefficient}$$
(25)

Water from imported beverages: The estimation approach is the same as water by the domestic production of exported beverages, as described in Eq. 16.

Water in biomass products: Usually, the input of biomass products by domestic extraction16 has been measured in fresh weight, but their corresponding output29 by sewage sludge, composting, etc., are in dry weight, leading to an imbalance in water weight. The water weight in biomass products is calculated by multiplying their domestic extraction amount in fresh weight (WBiomass) by a parameter of moisture content at harvest (PMoisture content), as shown in Eq. 26. The values of PMoisture content by biomass products are presented in Table 5.Table 5 The moisture content at harvest for each biomass product4.Full size table

$${W}_{Water}={W}_{Biomass}times {P}_{Moisturecontent}$$
(26)
Material flow quantificationThe above attempts have quantified material inputs and outputs by flows and presented a detailed profile of material utilisation for each material in China’s economy. In order to depict the economy in a more general way, EW-MFA indicators are assessed by aggregating flows by materials or periods as below.

Domestic extraction (DE): is referred to as natural materials that are extracted from the domestic environment and are used in the domestic economy, i.e., the total input of natural materials by extraction.

Domestic processed output (DPO): is referred to as materials that are released to the domestic environment after being processed in the domestic economy, i.e., the total output of processed materials by release.

Import (IM): is referred to as all goods (in the form of raw materials, semi-finished materials, and final products) that originated from other economies and are further used in the domestic economy. It is calculated as the sum of all imported goods.

Export (EX): is referred to as all goods that originated from the domestic economy and are transported to other economies to be used. It is calculated as the sum of all exported goods.

Domestic material input (DMI): is referred to as materials that originated from the domestic environment by extraction and other economies and are available (to be used or to be stored) for the domestic economy. It is calculated as the sum of DE plus IM, as shown in Eq. 27.$$DMI=DE+IM$$
(27)

Domestic material consumption (DMC): is referred to as materials that are directly used in the domestic economy after parts of them are exported to other economies. It is calculated as the difference between DMI and EX.

Physical trade balance (PTB): is referred to as a surplus or deficit of materials for the domestic economy. It is calculated as the difference between IM and EX.

Net additions to stock (NAS): is referred to as materials that remain in the domestic economy. It is calculated by taking BI items into account, as shown in Eq. 28.

$$NAS=DMC+B{I}_{in}-DPO-B{I}_{out}$$
(28) More

Hwang, H. Y. et al. Effect of cover cropping on the net global warming potential of rice paddy soil. Geoderma 292, 49–58 (2017).ADS 
CAS 
Article 

Google Scholar 
IPCC. Climate change 2013: The physical science basis. The Working Group I contribution to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC, 2013).
Google Scholar 
Cardoso, R. M., Soares, P. M. M., Lima, D. C. A. & Miranda, P. M. A. Mean and extreme temperatures in warming climate: EURO CORDEX and WRF regional climate high-resolution projection for Portugal. Clim. Dyn. 52, 129–157 (2019).Article 

Google Scholar 
Ding, T., Gao, H. & Li, W. J. Extreme high-temperature event in southern China in 2016 and the possible role of cross-equatorial flows. Int. J. Climatol. 38, 3579–3594 (2018).Article 

Google Scholar 
Escalas, A. et al. Functional diversity and redundancy across fish gut, sediment, and water bacterial communities. Environ. Microbiol. 19, 3268–3282 (2017).Article 

Google Scholar 
Philippot, L. et al. Loss in microbial diversity affects nitrogen cycling in soil. ISME J. 7, 1609–1619 (2013).CAS 
Article 

Google Scholar 
Li, Y. B. et al. Serratia spp. Are responsible for nitrogen fixation fueled by As(III) oxidation, a novel biogeochemical process identified in mine tailings. Environ. Sci. Technol 56, 2033–2043 (2022).ADS 
Article 

Google Scholar 
Jia, M., Gao, Z. W., Gu, H. J., Zhao, C. Y. & Han, G. D. Effects of precipitation change and nitrogen addition on the composition, diversity, and molecular ecological network of soil bacterial communities in a desert steppe. PLoS ONE 16, e0248194. https://doi.org/10.1371/journal.pone.0248194 (2021).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Waghmode, T. R. et al. Response of nitrifier and denitrifier abundance and microbial community structure to experimental warming in an agricultural ecosystem. Front. Microbiol. 9, 474. https://doi.org/10.3389/fmicb.2018.00474 (2018).Article 
PubMed 
PubMed Central 

Google Scholar 
Hu, Y. L., Wang, S., Niu, B., Chen, Q. & Zhang, G. Effect of increasing precipitation and warming on microbial community in Tibetan alpine steppe. Environ. Res. 189, 109917. https://doi.org/10.1016/j.envres.2020.109917 (2020).CAS 
Article 
PubMed 

Google Scholar 
Li, H. et al. Responses of soil bacterial communities to nitrogen deposition and precipitation increment are closely linked with aboveground community variation. Microb. Ecol. 71, 974–989 (2016).CAS 
Article 

Google Scholar 
Wang, H. et al. Experimental warming reduced topsoil carbon content and increased soil bacterial diversity in a subtropical planted forest. Soil Biol. Biochem. 133, 155–164 (2019).CAS 
Article 

Google Scholar 
Haumann, F. A., Gruber, N. & Münnich, M. Sea-Ice Induced Southern Ocean Subsurface Warming and Surface Cooling in a Warming Climate. AGU Advances 1, e2019AV000132. https://doi.org/10.1029/2019AV000132 (2020).ADS 
Article 

Google Scholar 
Ji, F., Wu, Z. H., Huang, J. P. & Chassignet, E. P. Evolution of land surface air temperature trend. Nat. Clim. Chang. 4, 462–466 (2014).ADS 
Article 

Google Scholar 
Sabri, N. S. A., Zakaria, Z., Mohamad, S. E., Jaafar, A. B. & Hara, H. Importance of soil temperature for the growth of temperate crops under a tropical climate and functional role of soil microbial diversity. Microbes Environ. 33, 144–150 (2018).Article 

Google Scholar 
McGrady-Steed, J. & Morin, P. T. Biodiversity, density compensation, and the dynamics of populations and functional groups. Ecology 81, 361–373 (2000).Article 

Google Scholar 
Jiang, L. Density compensation can cause no effect of biodiversity on ecosystem function. Oikos 116, 324–334 (2007).Article 

Google Scholar 
Faust, K. & Raes, J. Microbial interactions: From networks to models. Nat. Rev. Microbiol. 10, 538. https://doi.org/10.1038/nrmicro283 (2012).CAS 
Article 
PubMed 

Google Scholar 
Gao, X. X. et al. Revegetation significantly increased the bacterial-fungal interactions in different successional stages of alpine grasslands on the Qinghai-Tibetan Plateau. CATENA 205, 105385. https://doi.org/10.1016/j.catena.2021.105385 (2021).CAS 
Article 

Google Scholar 
Morriën, E. et al. Soil networks become more connected and take up more carbon as nature restoration progresses. Nat. Commun. 8, 14349. https://doi.org/10.1038/ncomms14349 (2017).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Banerjee, S. et al. Agricultural intensification reduces microbial network complexity and the abundance of keystone taxa in roots. ISME J. 13, 1722–1736 (2019).Article 

Google Scholar 
Pržulj, N. & Malod-Dognin, N. Network analytics in the age of big data. Science 353, 123–124 (2016).ADS 
Article 

Google Scholar 
Ratzke, C., Barrere, J. M. R. & Gore, J. Strength of species interactions determines biodiversity and stability in microbial communities. Nat. Ecol. Evol. 4, 376–383 (2020).Article 

Google Scholar 
Fuhrman, J. A. Microbial community structure and its functional implications. Nature 45, 193–199 (2009).ADS 
Article 

Google Scholar 
Zhao, M. X., Cong, J., Cheng, J. M., Qi, Q. & Zhang, Y. G. Soil microbial community assembly and interactions are constrained by nitrogen and phosphorus in broadleaf forests of southern China. Forest 11, 285. https://doi.org/10.3390/f11030285 (2020).Article 

Google Scholar 
Wan, X. L. et al. Biogeographic patterns of microbial association networks in paddy soil within Eastern China. Soil Biol. Biochem. 142, 07696. https://doi.org/10.1016/j.soilbio.2019.107696 (2020).CAS 
Article 

Google Scholar 
Yuan, M. M., Guo, X., Wu, L., Zhang, Y. & Zhou, J. Climate warming enhances microbial network complexity and stability. Nat. Clim. Change 11, 343–348 (2021).ADS 
Article 

Google Scholar 
Lassaletta, L. et al. Food and feed trade as a driver in the global nitrogen cycle: 50-year trends. Biogeochemistry 11, 225–241 (2014).Article 

Google Scholar 
Phoenix, G. K. et al. Impacts of atmospheric nitrogen deposition: Responses of multiple plant and soil parameters across contrasting ecosystems in long-term field experiments. Glob. Change Biol. 18, 1197–1215 (2012).ADS 
Article 

Google Scholar 
Nakaji, T., Fukami, M., Dokiya, Y. & Izuta, T. Effects of high nitrogen load on growth, photosynthesis and nutrient status of Cryptomeria japonica and Pinus densiflora seedlings. Trees-Struct. Funct. 15, 453–461 (2001).CAS 
Article 

Google Scholar 
Wang, H. Y. et al. Reduction in nitrogen fertilizer use results in increased rice yields and improved environmental protection. Int. J. Agric. Sustain. 15, 681–692 (2017).Article 

Google Scholar 
Zhou, X. G. & Wu, F. Z. Land-use conversion from open field to greenhouse cultivation differently affected the diversities and assembly processes of soil abundant and rare fungal communities. Sci. Total Environ. 788, 147751. https://doi.org/10.1016/j.scitotenv.2021.147751 (2021).ADS 
CAS 
Article 
PubMed 

Google Scholar 
Guo, H. et al. Long-term nitrogen & phosphorus additions reduce soil microbial respiration but increase its temperature sensitivity in a Tibetan alpine meadow. Soil Biol. Biochem. 113, 26–34 (2017).CAS 
Article 

Google Scholar 
Zhang, C. et al. Effects of simulated nitrogen deposition on soil respiration components and their temperature sensitivities in a semiarid grassland. Soil Biol. Biochem. 75, 113–123 (2014).CAS 
Article 

Google Scholar 
Zhang, J. J. et al. Different responses of soil respiration and its components to nitrogen and phosphorus addition in a subtropical secondary forest. For. Ecosyst. 8, 37. https://doi.org/10.1186/s40663-021-00313-z (2021).Article 

Google Scholar 
Norse, D. & Ju, X. T. Environmental costs of China’s food security. Agric. Ecosyst. Environ. 209, 5–14 (2015).Article 

Google Scholar 
Xu, H. F., Du, H., Zeng, F. P., Song, T. Q. & Peng, W. X. Diminished rhizosphere and bulk soil microbial abundance and diversity across succession stages in Karst area, southwest China. Appl. Soil Ecol. 158, 103799. https://doi.org/10.1016/j.apsoil.2020.103799 (2020).Article 

Google Scholar 
Li, Y. B. et al. Arsenic and antimony co-contamination influences on soil microbial community composition and functions: Relevance to arsenic resistance and carbon, nitrogen, and sulfur cycling. Environ. Int. 153, 106522. https://doi.org/10.1016/j.envint.2021.106522 (2021).CAS 
Article 
PubMed 

Google Scholar 
Zhou, J. & Fong, J. J. Strong agricultural management effects on soil microbial community in a non-experimental agroecosystem. Appl. Soil Ecol. 165, 103970. https://doi.org/10.1016/j.apsoil.2021.103970 (2021).Article 

Google Scholar 
Bárcenas-Moreno, G., Gómez-Brandón, M., Rousk, J. & Bååth, E. Adaptation of soil microbial communities to temperature: Comparison of fungi and bacteria in a laboratory experiment. Glob. Chang. Biol. 15, 2950–2957 (2009).ADS 
Article 

Google Scholar 
Tan, E. H., Zou, W., Zheng, Z., Yan, X. & Kao, S. J. Warming stimulates sediment denitrification at the expense of anaerobic ammonium oxidation. Nat. Clim. Change 10, 349–355 (2020).ADS 
CAS 
Article 

Google Scholar 
Supramaniam, Y., Chong, C. W., Silvaraj, S. & Tan, K. P. Effect of short term variation in temperature and water content on the bacterial community in a tropical soil. Appl Soil Ecol. 107, 279–289 (2016).Article 

Google Scholar 
Zhu, Y. Z., Li, Y. Y., Zheng, N. G., Chapman, S. J. & Yao, H. Y. Similar but not identical resuscitation trajectories of the soil microbial community based on either DNA or RNA after flooding. Agronomy 10, 502. https://doi.org/10.3390/agronomy10040502 (2020).CAS 
Article 

Google Scholar 
Donhauser, J., Qi, W., Bergk-Pinto, B. & Frey, B. High temperatures enhance the microbial genetic potential to recycle C and N from necromass in high-mountain soils. Glob. Chang. Biol. 27, 1365–1386 (2021).ADS 
Article 

Google Scholar 
Santoyo, G., Hernandez-Pacheco, C., Hernandez-Salmeron, J. & Hernandez-Leon, R. The role of abiotic factors modulating the plant-microbe-soil interactions: Toward sustainable agriculture. A review. Span. J. Agric. Res. 15, e03R01-e11. https://doi.org/10.5424/sjar/2017151-9990 (2017).Article 

Google Scholar 
Lefcheck, J. S. et al. Biodiversity enhances ecosystem multifunctionality across trophic levels and habitats. Nat. Commun. 6, 6936. https://doi.org/10.1038/ncomms7936 (2015).ADS 
CAS 
Article 
PubMed 

Google Scholar 
Cardinale, B. J. et al. Corrigendum: Biodiversity loss and its impact on humanity. Nature 486, 59–67 (2012).ADS 
CAS 
Article 

Google Scholar 
Ma, B., Wang, H., Dsouza, M., Lou, J. & Xu, J. Geographic patterns of co-occurrence network topological features for soil microbiota at continental scale in eastern China. ISME J. 10, 1891–1901 (2016).CAS 
Article 

Google Scholar 
Trivedi, C. et al. Losses in microbial functional diversity reduce the rate of key soil processes. Soil Biol. Biochem. 135, 267–274 (2019).CAS 
Article 

Google Scholar 
Melanie, K. et al. Effects of season and experimental warming on the bacterial community in a temperate mountain forest soil assessed by 16S rRNA gene pyrosequencing. FEMS Microbiol. Ecol. 82, 551–562 (2012).Article 

Google Scholar 
Zheng, H. F., Liu, Y., Chen, Y., Zhang, J. & Chen, Q. Short-term warming shifts microbial nutrient limitation without changing the bacterial community structure in an alpine timberline of the eastern Tibetan Plateau. Geoderma 360, 113985. https://doi.org/10.1016/j.geoderma.2019.113985 (2020).ADS 
CAS 
Article 

Google Scholar 
Finlay, B. J. & Cooper, J. L. Microbial diversity and ecosystem function. CEH Integrating Fund second progress report to the Director, Centre for Ecology and Hydrology Nov 1996–Sept (1997).Xing, X. Y. et al. Warming shapes nirS- and nosZ-type denitrifier communities and stimulates N2O emission in acidic paddy soil. Appl. Environ. Microbiol. 87, e02965-e3020. https://doi.org/10.1128/AEM.0296520 (2021).CAS 
Article 
PubMed Central 

Google Scholar 
Lin, Y. T., Whitman, W. B., Coleman, D. C., Jien, S. H. & Chiu, C. Y. Soil bacterial communities at the treeline in subtropical alpine areas. CATENA 201, 105205. https://doi.org/10.1016/j.catena.2021.105205 (2021).CAS 
Article 

Google Scholar 
Wang, J. C. et al. Impacts of inorganic and organic fertilization treatments on bacterial and fungal communities in a paddy soil. Appl. Soil Ecol. 112, 42–50 (2017).Article 

Google Scholar 
Chacón, J. M., Shaw, A. K. & Harcombe, W. R. Increasing growth rate slows adaptation when genotypes compete for diffusing resources. PLoS Comput. Biol. 16, e1007585. https://doi.org/10.1371/journal.pcbi.1007585 (2020).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Hartley, I. P., Hopkins, D. W., Garnett, M. H., Sommerkorn, M. & Wookey, P. A. Soil microbial respiration in arctic soil does not acclimate to temperature. Ecol. Lett. 11, 1092–1100 (2008).Article 

Google Scholar 
Baath, E. Growth rates of bacterial communities in soils at varying pH: A comparison of the thymidine and leucine incorporation techniques. Microb. Ecol. 36, 316–327 (1998).CAS 
Article 

Google Scholar 
Qin, H. L. et al. Soil moisture and activity of nitrite- and nitrous oxide-reducing microbes enhanced nitrous oxide emissions in fallow paddy soils. Biol. Fertil. Soils 56, 53–67 (2020).CAS 
Article 

Google Scholar 
Chen, Z. et al. Impact of long term fertilization on the composition of denitrifier communities based on nitrite reductase analyses in a paddy soil. Microb. Ecol. 60, 850–861 (2010).CAS 
Article 

Google Scholar 
Wei, G. S. et al. Similar drivers but different effects lead to distinct ecological patterns of soil bacterial and archaeal communities. Soil Biol. Biochem. 144, 107759. https://doi.org/10.1016/j.soilbio.2020.107759 (2020).CAS 
Article 

Google Scholar 
Bastian, F., Bouziri, L., Nicolardot, B. & Ranjard, A. L. Impact of wheat straw decomposition on successional patterns of soil microbial community structure. Soil Biol. Biochem. 41, 262–275 (2009).CAS 
Article 

Google Scholar 
Levins, R. Evolution in Changing Environments: Some Theoretical Explorations (Princeton University Press, 1968).Book 

Google Scholar  More

IPBES. Global assessment report on biodiversity and ecosystem services of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services. Zenodo https://doi.org/10.5281/zenodo.5657041 (2019).Laurance, W. F. et al. A global strategy for road building. Nature 513, 229–232 (2014).ADS 
CAS 
Article 

Google Scholar 
Ibisch, P. L. et al. A global map of roadless areas and their conservation status. Science 354, 1423–1427 (2016).ADS 
CAS 
Article 

Google Scholar 
Forman, R. T. T., Sperling, D. & Bissonette, J. A. Road Ecology: Science and Solutions. (Island Pr, 2003).van der Ree, R., Smith, D. J. & Grilo, C. Handbook of Road Ecology. (John Wiley & Sons, 2015).Laender. Hunting Statistics. Game casualties 2017/2018: furred game (red deer, roe deer, chamois, moufflon) https://www.statistik.at/web_en/statistics/Economy/agriculture_and_forestry/livestock_animal_production/hunting/index.html (2018).Steiner, W., Leisch, F. & Hacklander, K. A review on the temporal pattern of deer-vehicle accidents: Impact of seasonal, diurnal and lunar effects in cervids. Accident; analysis and prevention 66, (2014).Kioko, J. et al. Driver knowledge and attitudes on animal vehicle collisions in Northern Tanzania. TROPICAL CONSERVATION SCIENCE 8, 352–366 (2015).Article 

Google Scholar 
Bíl, M., Andrášik, R. & Janoška, Z. Identification of hazardous road locations of traffic accidents by means of kernel density estimation and cluster significance evaluation. Accident Analysis & Prevention 55, 265–273 (2013).Article 

Google Scholar 
Page, Y. A statistical model to compare road mortality in OECD countries. Accident Analysis and Prevention 33, 371–385 (2001).CAS 
Article 

Google Scholar 
Teixeira, F. Z. et al. Are Road-kill Hotspots Coincident among Different Vertebrate Groups? Oecologia Australis 17, 36–47 (2017).Article 

Google Scholar 
Canova, L. & Balestrieri, A. Long-term monitoring by roadkill counts of mammal populations living in intensively cultivated landscapes. Biodivers Conserv https://doi.org/10.1007/s10531-018-1638-3 (2018).Brehme, C. S., Hathaway, S. A. & Fisher, R. N. An objective road risk assessment method for multiple species: ranking 166 reptiles and amphibians in California. Landscape Ecol 33, 911–935 (2018).Article 

Google Scholar 
Heigl, F. et al. Comparing Road-Kill Datasets from Hunters and Citizen Scientists in a Landscape Context. Remote Sensing 8, (2016).Heigl, F., Horvath, K., Laaha, G. & Zaller, J. G. Amphibian and reptile road-kills on tertiary roads in relation to landscape structure: using a citizen science approach with open-access land cover data. BMC Ecol 17, 24 (2017).Article 

Google Scholar 
Dörler, D. & Heigl, F. A decrease in reports on road-killed animals based on citizen science during COVID-19 lockdown. PeerJ 9, e12464 (2021).Article 

Google Scholar 
Peer, M. et al. Predicting spring migration of two European amphibian species with plant phenology using citizen science data. Sci Rep 11, 21611 (2021).ADS 
CAS 
Article 

Google Scholar 
Schwartz, A. L. W. UK Roadkill Records. The Global Biodiversity Information Facility https://doi.org/10.15468/r3xakd (2018).Lin, T. The Taiwan Roadkill Observation Network Data Set. Version 1.3. The Global Biodiversity Information Facility https://doi.org/10.15468/cidkqi (2018).Chandler, M. et al. Contribution of citizen science towards international biodiversity monitoring. Biological Conservation 213, 280–294 (2017).Article 

Google Scholar 
Périquet, S., Roxburgh, L., le Roux, A. & Collinson, W. J. Testing the Value of Citizen Science for Roadkill Studies: A Case Study from South Africa. Front. Ecol. Evol. 6, (2018).Abra, F. D., Huijser, M. P., Pereira, C. S. & Ferraz, K. M. P. M. B. How reliable are your data? Verifying species identification of road-killed mammals recorded by road maintenance personnel in São Paulo State, Brazil. Biological Conservation 225, 42–52 (2018).Article 

Google Scholar 
Bíl, M., Kubeček, J., Sedoník, J. & Andrášik, R. Srazenazver.cz: A system for evidence of animal-vehicle collisions along transportation networks. Biological Conservation 213, 167–174 (2017). Part A.Article 

Google Scholar 
Vercayie, D. & Herremans, M. Citizen science and smartphones take roadkill monitoring to the next level. Nature Conservation 11, 29–40 (2015).Article 

Google Scholar 
Waetjen, D. P. & Shilling, F. M. Large Extent Volunteer Roadkill and Wildlife Observation Systems as Sources of Reliable Data. Front. Ecol. Evol. 5, (2017).Shilling, F. M., Perkins, S. E. & Collinson, W. Wildlife/Roadkill Observation and Reporting Systems. in Handbook of Road Ecology 492–501 (John Wiley & Sons, 2015).Eitzel, M. V. et al. Citizen Science Terminology Matters: Exploring Key Terms. Citizen Science: Theory and Practice 2, 1–20 (2017).
Google Scholar 
Haklay, M. et al. Contours of citizen science: a vignette study. Royal Society Open Science 8, 202108 (2021).ADS 
Article 

Google Scholar 
Heigl, F., Kieslinger, B., Paul, K. T., Uhlik, J. & Dörler, D. Opinion: Toward an international definition of citizen science. PNAS 116, 8089–8092 (2019).CAS 
Article 

Google Scholar 
Heigl, F. et al. Quality Criteria for Citizen Science Projects on Österreich forscht | Version 1.1. Open Science Framework https://doi.org/10.17605/OSF.IO/48J27 (2018).Heigl, F. et al. Co-Creating and Implementing Quality Criteria for Citizen Science. Citizen Science: Theory and Practice 5, 23 (2020).
Google Scholar 
Heigl, F. & Zaller, J. G. Using a Citizen Science Approach in Higher Education: a Case Study reporting Roadkills in Austria. Human Computation 1, (2014).University of Natural Resources and Life Sciences, Vienna. Roadkill, The Global Biodiversity Information Facility, https://doi.org/10.15468/ejb47y (2021).Heigl, F. & Roadkill Community. Roadkill Dataset 2014-2020 Quality level 2, Zenodo, https://doi.org/10.5281/zenodo.5878813 (2022).August, T. A. et al. Citizen meets social science: predicting volunteer involvement in a global freshwater monitoring experiment. Freshwater Science 38, 321–331 (2019).Article 

Google Scholar 
IUCN. The IUCN Red List of Threatened Species. Version 2021-3. IUCN Red List of Threatened Species https://www.iucnredlist.org/en (2021). More

Taxonomy, variation, and biochronologyThe fossils described herein represent one of the most valuable and best-preserved samples of “Dama-like” deer from the European Early Pleistocene. The systematics of these forms has been essentially based on the morphology of the antlers and teeth, with less attention paid to the skull (due to the rarity of well-preserved finds) and postcranial bones.The Pantalla sample shows a combination of characters allowing an unambiguous attribution to ‘P.’ nestii, a species reported confidently so far in the early Late Villafranchian of Italy (several sites) and in the Georgian Homo-bearing locality of Dmanisi (Supplementary Table S1). Based on the literature6,8,12,38, these characters include: four-pointed antler with elongated, slender, and tubular beam; basal tine branching off at a certain distance from the burr forming an acute angle; well-developed middle tine; terminal bifurcation oriented normal to the sagittal plane; cranium with large orbits, preorbital fossae, and ethmoidal vacuities; relatively elongated neurocranium with flat parietals; caudally-oriented pedicles; molarized P2-P3; presence of cingula in upper molars; enlarged i1; un-molarized p4. However, some characters observed in the Pantalla specimens (e.g., rostral edge of the orbit reaching the level of M2; elongated metapodials) do not fit the revised diagnosis of ‘P.’ nestii by Croitor12. The latter author considers nestii as the earliest species of the genus Cervus based on similarities with the extant red deer especially in cranial morphology12,22,23. However, in our opinion, his conclusions are biased by relying mostly on the skull IGF 243 of ‘P.’ nestii from Upper Valdarno6,8, which is heavily deformed and belongs to a juvenile individual (see below for details on ontogenetic variation in ‘Pseudodama’).A broader look at the entire record of ‘P.’ nestii reveals that this species displays a mosaic of characters between Dama and Cervus, but also that the shared characters with Dama are prevalent (as already pointed out by Azzaroli8). The Pantalla sample allows to substantiate these conclusions very well. Our CT-based comparisons between the crania from Pantalla and those of extant red deer and fallow deer (Fig. 3) highlight some morphological similarities with the former, including a relatively longish neurocranium with steep forehead and deep preorbital fossa. On the other hand, ‘P.’ nestii from Pantalla clearly shows Dama-like cranial characters, such as a marked interfrontal crest, horizontal zygomatic arch, high maxilla below the orbit, muzzle more inclined ventrally and less cylindrical in overall shape, sub-horizontal upper cheek tooth row (i.e., the occlusal margin of the row is approximately straight in buccal view), apical surface of the pedicle more inclined dorsocaudally, and overall morphology of the antlers, which in rostral view diverge, rather than converge as in the red deer (Fig. 2).Likewise, the teeth from Pantalla, have a mixture of Dama and Cervus characters although the former are prevalent. All the premolar characters (the complete absence of a lingual grove on P4, the presence of a cingulum on the distolingual wall of P4, the presence of a small paraconid in p2, the entoconid more aligned with the mesiodistal axis in p3-p4, and a weak mesial cingulum on p4) and most of the lower molar characters are Dama-like. The upper molar features are instead more reminiscent of Cervus being either intermediate between the morphology of the latter and that of extant Dama or even matching Cervus (see Supplementary Table S6 and below).The postcranial remains from Pantalla appear more similar to Dama than to Cervus. Of the 23 morphological characters by Lister39 which are present in the preserved bones (axis, metacarpal, tibia, astragalus, calcaneum, cubo-navicular, metatarsal, phalanx I, and phalanx II), 21 scores as fallow deer and only two as red deer (details in Supplementary Table S7).A mixed character suite between Dama and Cervus are revealed also by our palaeoneurological analysis. The brain of ‘P.’ nestii shows Dama-like size and Cervus-like morphology with a prominent cerebellum and a dorsoventrally flattened cerebrum. The latter character is clearly noticeable in ‘Pseudodama’ and Eucladoceros, is less evident in extant Cervus, and is missing in Dama. The hypothesis that depressed and longish cerebra represent a primitive character in Cervini (at least in Pleistocene European forms) is supported by our preliminary data and agree with Azzaroli8.Most interestingly, the two crania from Pantalla actually show some remarkable morphological differences. The neurocranium of 337643 is more lengthened (i.e., more Cervus-like), albeit this shape might be taphonomically modified by the lateral compression of the specimen. This morphology fits that observed in some other ‘P.’ nestii specimens such as IGF 1403 from Olivola (Italy), while the relatively shorter and more rounded neurocranium of 337655 resembles that of other specimens such as IGF 1404 also from Olivola. Moreover, 337643 shows a stronger nuchal crest than 337655. These differences may be related to ontogenesis (see the advanced age of 337643 based on tooth wear). In several cervid species including fallow deer, aging leads to morphological changes in the neurocranium, which tends to elongate and flatten and shows a more developed nuchal crest, probably as a response to the support of larger and heavier antlers18,38. Similarly, in 337643, the pedicles are apparently closer to one another due to their thickening—an expected condition for an old individual as the distance between the pedicles tends to decrease with age8—and markedly shorter than wide. Our comparative data on European Dama-like deer show that the pedicle section can be highly variable both within and between species, although a general trend of laterolateral flattening (i.e., oval shape with major axis oriented anteroposteriorly) can be traced through time (Supplementary Fig. S3), probably as a result of the development of wide, laterally-projecting palmated antlers (in extant deer, D. dama is among those with the heaviest antlers relative to body size40,41). Therefore, the Pantalla sample on the one hand confirms the variation in cranial morphology already observed for ‘P.’ nestii6,8, on the other hand it supports the affinities between this species and the fallow deer. The presence of Cervus-like features especially in cranial morphology may be interpreted as plesiomorphic characters which, associated with some characters of the dentition and of the brain, suggest a basal position of ‘Pseudodama’ in the evolutionary history of the Cervini. This hypothesis may be tested in the future through phylogenetic analyses, currently made difficult by the lack of sufficiently well-preserved material of some species of ‘Pseudodama’ (e.g., ‘P.’ lyra, ‘P.’ perolensis).Compared with other specimens of ‘P.’ nestii6,8, the sample from Pantalla shows some plesiomorphic characters including a high ratio between the premolar and molar lengths, i.e., 0.77–0.82 (n = 2) for upper teeth (LP/LM) and 0.68–0.69 (n = 3) for lower teeth (Lp/Lm). These values are closer to the basal forms of ‘Pseudodama’, such as ‘P.’ lyra from Montopoli (LP/LM = 0.73, n = 1; Lp/Lm = 0.64, n = 2) and ‘P.’ rhenana from Saint Vallier (LP/LM = 0.75, n = 9; Lp/Lm = 0.68, n = 18; data from Valli42), than to ‘P.’ nestii from Olivola and Upper Valdarno (LP/LM = 0.72, n = 10; Lp/Lm = 0.63, n = 17). Other putatively plesiomorphic features of the sample from Pantalla are all those that approach it morphologically to Cervus (see Supplementary Table S6), i.e., the strong development of lingual conids and stylids in lower molars (Char. 439) and of buccal cones and styles in upper molars (Char. 139), the lack of a clear step between 2nd and 3rd lobe of m3 (Char. 1139), the strong lingual cingulum on upper molars (Char. 339), and the lack of the horizontal turning of the buccal columns of upper molars (Char. 439—the so-called buccal “cingulum”43). The strong lingual cingulum on upper molars is constantly present in the earliest species of the ‘Pseudodama’ group, ‘P.’ pardinensis9, and still present, although extremely rare, in ‘P.’ lyra from Montopoli, ‘P.’ rhenana from Saint Vallier and Senèze, ‘P.’ perolensis from Peyrolles, and ‘P.’ nestii from Olivola. However, this feature is back less rare in ‘P.’ nestii from Upper Valdarno and ‘P.’ farnetensis from Selvella, suggesting a certain polymorphism at this stage. The lack of buccal “cingulum” is a constant in the earliest ‘Pseudodama’ populations (Lower Valdarno, Saint Vallier, Senèze), the buccal “cingulum” appearing, although rare, in ‘P.’ perolensis from Peyrolles and ‘P.’ nestii from Olivola and Upper Valdarno but becoming more common only in later ‘P.’ farnetensis, ‘P.’ vallonnetensis, and constant in Dama.The above affinities between the Pantalla deer and the early representatives of ‘Pseudodama’ support the idea that the age of the assemblage may be close to the beginning of the Late Villafranchian (ca. 2.1–2.0 Ma), as already suggested based on the occurrence of Leptobos merlai44 and a primitive form of Equus stenonis35. Thus, the ‘P.’ nestii sample described herein may represent one of the earliest occurrences of the species in Europe.Palaeoecological and palaeoethological inferencesThe Pantalla sample is also noteworthy as it allows opening a window into the behaviour of these extinct deer. The anomalies found on the two male crania are probably the result of different traumas during their life.Deer are well known for the intense fights they engage in during the rutting season using their antlers, as a result of an escalation of a broad repertoire of threats and displays45. Mineralized antlers are solid structures able to withstand the vehemence of the fight46, whereas growing antlers are extremely fragile and any contact with a solid object may result in a serious injury47,48 that may jeopardize the bearer’s ability to compete with conspecifics and, consequently, its dominance status49. Accidents are inevitable in the life of a deer and, in case of the suffered damage not leading to the breakage of the growing beam and consequent loss of its distal part, the antler may continue its growth although, in case of a severe lesion, at a crooked angle45. Thus, if the antler was just cracked and the broken part was held together by the velvet and periosteum, with the blood supply still being guaranteed, the damaged beam would just present a conspicuous swelling around the area of fracture (i.e., a fracture callus)45,50 and a change in the axis of orientation. These features match those seen in the left beam of 337655, which shows a fracture callus between the basal and middle tines corresponding to a change in the orientation of the beam.The supernumerary tine of the right antler of the same individual can be interpreted as the result of a trauma, too. Considering the delicate nature of the growing antlers and the non-negligible risks of occurrence of an injury, it is safe to believe that the right antler has undergone a light traumatic event (most likely concerning the pedicle) at some early stage of its growth. In fact, it is known that limited injuries could result in the growth of supernumerary tines, even in atypical positions51, as it has been documented in other deer species (e.g., reindeer52, sambar53). It is therefore reasonable to hypothesize that both antler anomalies of 337655 derive from traumas suffered by the deer during the antler growth, when the velvet was still present. It is not known whether the two injuries happened at the same time or in two different events. In fact, it cannot even be said that the two events took place during the same season. While the breakage of the left beam must have occurred in the year of the animal’s death (i.e., during the velvet period preceding the period of hard antler in which the individual died), the development of the supernumerary tine on the right may be the result of a trauma suffered in a previous year. This is due to the fact that when unilateral trauma affects the generative region of the antler (i.e., the pedicle area), abnormalities such as supernumerary tines can reappear in next antler cycles even in more intensified forms54, as in the case of 337655 in which the extra-tine is extremely long.The bone anomaly on the right squamosal of 337643 is also likely the outcome of an injury. Although the external portion was artificially smoothed during the preparation of the specimen, the outer and inner morphology matches that of a callus related to the healing of a major lesion and probable intracranial abscessation. Post-traumatic inflammatory processes are known to cause erosion or pitting of cranial bones in deer55 and can be triggered by many factors (e.g., wounds and abrasions of the pedicle56), among which violent sexual competition among males with hard antlers is considered one of the most common55,57. The advanced healing of the injury shown by 337643 suggests that it was not the cause of death, but rather that the individual survived a long time after the trauma albeit with the brain partially compressed by the callus.The six mandibles recovered at Pantalla, all coming from the same bone accumulation hence reasonably referable to a single deer population, represent several age classes, from calves as young as a few months up to very old individuals (i.e., over 15 years; Supplementary Table S4). Unfortunately, no mandible can be safely associated with the two male crania, although 337631 may belong to the same individual as 337643 based on advanced wear and size. Interestingly, the three most significant cranial remains (crania 337655 and 337643 and frontal bone fragment with basal antler base 337625) belong to adult males, which probably died during the hard antler period (i.e., rutting season: 337655 and 337625) or shortly after (i.e., 337643). The absence of females (at least among the remains with certain sex attribution) contrasts with the population structure in the extant fallow deer, in which females represent on average 75% of the herd58. However, the relative abundance of males may increase up to 50% in the rutting season59,60. Therefore, in spite of the relatively low number of fossils available, based on the age and sex structure of the palaeopopulation and by analogy with the extant fallow deer, the most plausible hypothesis is that the Pantalla deer died during or immediately after the rutting season (Fig. 5).Figure 5Life appearance of ‘Pseudodama’ nestii represented during the rutting season. The reconstruction is based on the cranial and postcranial material from the Early Pleistocene of Pantalla (Italy) and on literature data. Artwork by D.A. Iurino.Full size image More

Experimental set-upFour vegetation species were selected: Spartina maritima, Salicornia europaea, Halimione portulacoides and Juncus maritimus. These species were chosen for a broad representation of the biomechanical properties and morphological characteristics of saltmarsh species42,43. Plants were collected in Cantabrian estuaries in late summer and early autumn (from early September to late October) during low tide (please refer to the “Methods” section). A total of 105 boxes were collected, of which 94 boxes were used to build a 9.05 m long and 0.58 m wide meadow in a flume (Fig. 1). Five boxes were used to directly estimate the meadow standing biomass in the field (Sample 1 in Table 1), leaving 6 extra boxes for possible contingencies.Figure 1(A) Shows a sketch of the experimental flume, where the vegetation box distribution in the 100% and 50% density cases is displayed in the two upper panels and a lateral view in the bottom panel. The green boxes indicate the vegetated area in each case. Free surface sensors are displayed by blue lines and numbers. (B) Shows the four species within the flume. From left to right: view of the Spartina sp. frontal edge, aerial view of Salicornia sp., frontal view of Juncus sp. and top view of the Halimione sp. rear edge.Full size imageTable 1 Standing biomass (g/m2) and plant height (m) for the four species.Full size tableExperiments were conducted in a flume 20.71 m long and 0.58 m wide at the University of Cantabria. The flume is equipped with a piston wave maker at its left end and a dissipation beach at the rear end. The 94 vegetation boxes used to create a meadow were introduced into the flume following the pattern shown in panel A of Fig. 1 to minimize any edge effects along the edges of the boxes. To ensure a smooth transition from the bottom of the channel to the vegetated area, two false bottoms were constructed with wood, and a thin sediment layer was glued to the wood to mimic the field roughness.Three meadow densities per species were considered. The meadow density directly determined in the field was chosen under the 100% density scenario. To consider a second meadow density, and therefore a second standing biomass value, plants were removed from half of the boxes following the pattern shown in Panel A of Fig. 1 to prevent creating preferential flow channels along the meadow. This case was considered the 50% density scenario. The study of these two biomass scenarios for each vegetation species is carried out with the aim of covering a wide range of standing biomass values, including low values that may be more representative of meadow winter conditions, thus facilitating the applicability of obtained results. Finally, a second cut was made, in which all plants were removed, resulting in the final scenario with a zero density. Plants were cut from above to avoid any damage along the meadow surface (as shown in Supplementary Fig. S2). In each cut, plants in 5 boxes along the leading edge and in 5 boxes at the center of the meadow were collected to quantify the standing biomass (Samples 2 and 3 for the first cut and Sample 4 and 5 for the second cut in Table 1). Therefore, the standing biomass could be monitored throughout the entire duration of the experiments, from the field until the second cut, when all plants were removed.Once located in the flume, the meadow was evaluated under regular and random wave conditions considering three water depths, i.e., h = 0.20, 0.30 and 0.40 m. Regular waves were generated using Stokes II-, III- and V-order and Cnoidal theories when applicable. Wave heights ranging from 0.05 to 0.15 m and wave periods varying between 1.5 and 4 s were considered. Random waves were generated using a Jonswap spectrum with a peak enhancement factor of 3.3, a significant wave height varying between 0.05 and 0.15 m and a peak wave period ranging from 1.8 to 4.8 s (please refer to Supplementary Table S1). Additionally, all wave conditions were considered under the zero-density scenario with bare soil for each species. The wave height evolution along the flume was recorded using 15 capacitive free surface gauges, as shown in Fig. 1 (please refer to Supplementary Table S2 for detailed coordinates).Meadow characteristics analysisThe characteristics of the vegetation meadows were analyzed by measuring the standing biomass throughout the full duration of the experiments and by measuring the individual plant height (please refer to the “Methods” section). The mean standing biomass value obtained for each species was considered the value associated with the 100% density scenario. Then, half of the standing biomass value was considered under the 50% density scenarios since half of the boxes was randomly cut, and the standing biomass values obtained after the second cut agreed with those obtained after the first cut and in the field, as indicated in Table 1. The plant height for each species was also measured (please refer to the “Methods” section), and the resultant mean value detailed in Table 1 was considered.Wave height attenuation analysisWave height attenuation analysis was performed following previous studies reported in the literature assessing the capacity by fitting a damping coefficient6,7,35,44. The18 formulation was used for regular waves, and that of19 was used for random waves (please refer to the “Methods” section). Cases with a zero density were also considered in this analysis to quantify the influence of bare soil friction by determining the corresponding damping coefficient, ({beta }_{B}). Consequently, β was obtained in the 100% and 50% density cases and the cases without vegetation (please refer to Supplementary Tables S3, S4 and S5 to find the obtained coefficients for all cases). This allowed the determination of a new damping coefficient isolating the effect of the standing biomass, ({beta }_{SB}), following24 (please refer to the “Methods” section). Figure 2 shows an example of wave height attenuation analysis for the four species and the different densities under wave condition JS07 (Supplementary Table S1).Figure 2Analysis of wave attenuation under wave condition JS07 for Spartina sp. 100% (S100), 50% (S050) and zero density (S000); Salicornia sp. 100% (L100), 50% (L050) and zero density (L000); Juncus sp. 100% (J100), 50% (J050) and zero density (J000); and Halimione sp. 100% (H100), 50% (H050) and zero density (H000). The damping coefficients for the bare soil cases, ({beta }_{B}), are displayed in blue. The damping coefficients for the 100% and 50% density cases, (beta ), are displayed in dark and light green, respectively. The damping coefficients obtained after subtracting the dissipation obtained in the bare soil cases, ({beta }_{SB}), are displayed in black and dark gray. 95% confidence interval is shown in brackets and correlation coefficient (({rho }^{2})) for each fit is also displayed.Full size imageThe damping coefficients for the bare soil cases shown in Fig. 2, ({beta }_{B}), are consistent with the soil properties observed in the field. Spartina sp. was collected in a muddy area, whereas the other three species were collected in areas with coarser sediments and exhibited a mixture of sand and mud. For all species, wave dissipation was significantly higher under the 100% density scenario than that under the 50% density cases, as expected, highlighting the importance of the standing biomass in wave energy dissipation. It was also observed that bottom friction-induced dissipation plays a more important role for the pioneer species, i.e., Spartina sp. and Salicornia sp., than for the upper marsh species, i.e., Juncus sp. and Halimione sp., which can dissipate wave energy to a greater extent.The importance of wave parameters in the resultant wave attenuation has been highlighted by several works in the literature. Therefore, not only vegetation characteristics but also incident wave conditions determine the coastal protection capacity. Figure 3 shows a comparison of the obtained wave height attenuation due to Halimione sp. under the different wave conditions.Figure 3Analysis of wave attenuation under the different irregular wave conditions for the Halimione sp. 100% (H100) and zero-density (H000) cases. The top panel shows two cases with different h but equal Hs and Tp values (JS01 and JS08), the middle panel shows two cases with different Tp but equal h and Hs values (JS10 and JS11), and the bottom panel shows two cases with different Hs but equal h and Tp values (JS09 and JS12). 95% confidence interval is shown in brackets and correlation coefficient (({rho }^{2})) for each fit is also displayed.Full size imageThe top panel in Fig. 3 shows two cases where Hs and Tp are equal, i.e., JS01 and JS08 in Supplementary Table S1, and two water depths are considered, namely, h = 0.2 and 0.3 m. As can be observed, wave damping is higher for the smallest water depth, where most of the water column is covered by vegetation since the mean vegetation height for Halimione sp. reaches 0.187 m (Table 1). The importance of the water depth with respect to the plant height in terms of wave height attenuation has been reported by several authors44,45,46 who have highlighted this aspect based on the submergence ratio, i.e., the plant height divided by the water depth, revealing higher attenuation at lower submergence ratios on a consistent basis. Bottom friction attenuation is also higher for the smallest water depth, as expected.The middle panel of Fig. 3 shows two cases with equal h and Hs but different Tp values, namely, JS10 and JS11 in Supplementary Table S1. Wave height attenuation is higher for the shortest wave period, as well as the damping produced by bottom friction. This is in line with previous studies, such as35 and44, who conducted experiments involving simulated and real saltmarshes, respectively. Finally, the bottom panel of Fig. 3 shows two cases with different Hs but equal h and Tp values, i.e., JS09 and JS12 in Supplementary Table S1. As widely reported in the literature, e.g.,7,47,48, wave height attenuation increases with the wave height, as shown in the bottom panel of Fig. 3. Bottom friction also increases with the wave height, as expected.A set of damping coefficients was obtained via the 288 tests conducted in the laboratory, 144 tests involving regular waves and 144 tests involving random waves. Additionally, in all cases, the damping coefficient considering the isolated effect of the standing biomass, ({beta }_{SB}), was determined. The relationship of these damping coefficients to the measured standing biomass is explored in the next section with the aim of establishing a new relationship to estimate the wave damping effect of the different saltmarsh species based on the standing biomass, without the need for data fitting.Wave damping coefficient as a function of the standing biomassThe mean standing biomass obtained for the different species, Table 1, is considered here to analyze the relationship with the wave damping coefficients obtained by fitting18 formulation to wave heights measured along the meadow for regular waves and19 formulation for random waves. The plant height was highly variable among the different species (Table 1), ranging from 0.170 m for Spartina sp. to 0.714 m for Juncus sp. Then, some species were submerged at all tested water depths, while other species remained above water in all tests. In the latter cases, there remained a portion of each plant above the water level, thus not contributing to wave attenuation. To consider the actual interaction between the standing biomass and flow conditions and assuming a uniform vertical distribution, the effective standing biomass, (ESB), can be defined as follows:$$ESB=DryWeight*frac{minleft{{h}_{v},hright}}{{h}_{v}}$$
(1)
where (DryWeight) denotes the measured dry weight for each species (g/m2), ({h}_{v}) is the mean plant height and (h) is the water depth. Additionally, in the submerged cases, the same (ESB) value will impact flow differently depending on the submergence ratio, (SR), as defined in Eq. (2). To consider this effect, the standing biomass ratio, (SBR) in Eq. (3), can be defined as follows:$$SR=frac{{h}_{v}}{h}, ;;where ;; SR=1 ;;for ;;{h}_{v} >h$$
(2)
$$SBR=ESB*SR$$
(3)
Figure 4 shows the relationship between (SBR) and the measured wave damping coefficient, (beta ). The results for regular and random waves are displayed for each water depth, and a linear fit was found under each condition.Figure 4Wave damping coefficient, (beta ), as a function of the standing biomass ratio, (SBR), under all regular (left panels) and random (right panels) wave conditions. Each panel shows the wave trains assessed at each water depth, h = 0.20, 0.30 and 0.40 m. The results for the 100% density case are marked with circles and those for the 50% density case are marked with squares. The linear fitting results obtained under each wave condition are also displayed.Full size imageUnder each wave condition, a linear fitting relationship between (beta ) and (SBR) was obtained for the eight (SBR) values, as shown in Fig. 4. For similar (SBR) values, the highest (beta ) values were consistently obtained at the smallest water depth, highlighting the notable influence of this parameter on the obtained wave attenuation. Following previous works, such as those of24 and25, who considered the vegetation submerged solid volume fraction to estimate the resulting wave attenuation and established a common relationship for different water depths, the volumetric standing biomass, (VSB), can be defined as follows:$$VSB= SBR*frac{1}{h}$$
(4)
(VSB) is expressed in units of g/m3, which is the weight per unit volume. Exploring the relationship of (beta ) with this new parameter, it was found that the results for the three water depths could be fitted with a single linear relationship, as shown in Fig. 5. However, despite the linear trend observed in Fig. 5, notable data scatter was observed for each (VSB) value. Each of these groups corresponds to a certain water depth and (SBR) value, which were determined under different wave heights and wave periods.Figure 5Wave damping coefficient, (beta ), as a function of the volumetric standing biomass, (VSB), under all regular (top panel) and random (bottom panel) wave conditions. The obtained linear fitting results are displayed in both panels. 95% confidence interval is shown in brackets and correlation coefficient (({rho }^{2})) for each fit is also displayed.Full size imageFinally, to account for the characteristics of the incident wave conditions, including the wave height and period, two nondimensional parameters were considered. The first parameter, considering the wave height, is the relative wave height, defined as the ratio of the incident wave height to the water depth, (H/h). Previous studies have highlighted the importance of this parameter in the resultant wave attenuation (e.g.24,44). Under random wave conditions, the considered wave height is ({H}_{rms}), according to wave attenuation analysis. The second parameter, considering the effect of the different wave periods and the importance of the number of wave lengths inside the vegetation length49, is the relative meadow length, defined as the ratio of the meadow length to the wave length, ({L}_{v}/L). To ensure consistency with the above wave attenuation analysis, in which the wave damping amount per unit length was obtained, the unit meadow length was considered here. Thus, the hydraulic standing biomass, (HSB), can be defined as:$$HSB=VSB*frac{H}{h}*frac{{L}_{v}}{L}$$
(5)
Figure 6 shows the relationship obtained between (beta ) and this new variable under all regular and random conditions following the linear fitting relationship of (beta =A*HSB+B), where (A) and (B) are fitting constants with units of (g/m2)−1 and m−1, respectively.Figure 6Wave damping coefficient, (beta ), as a function of the hydraulic standing biomass, (HSB), under all regular (top panel) and random (bottom panel) wave conditions. Both panels show linear fitting results obtained without considering the saturation point, indicated by the black solid line, and those obtained considering the saturation point, indicated by the gray solid line. The black dashed line indicates the saturation point. 95% confidence interval is shown in brackets and correlation coefficient (({rho }^{2})) for each fit is also displayed.Full size imageThe linear fitting results obtained between (beta ) and (HSB) under regular and random wave conditions are shown in Fig. 6 as solid black lines and expressed as Eqs. (6) and (7), respectively, where values between brackets are the 95% confidence interval for each coefficient.$$beta =9.206cdot {10}^{-4} left(9.006cdot {10}^{-5}right)*HSB+0.103 (0.021)$$
(6)
$$beta =1.192 cdot {10}^{-3} left(9.124 cdot {10}^{-5}right)*HSB+0.071 (0.016)$$
(7)
The inclusion of incident wave condition characteristics reduces the resulting data scatter, highlighting the role of the wave height and period in the obtained wave attenuation, as described in the previous section. An interesting aspect observed in Fig. 6 is that the four cases with the highest wave damping coefficients yielded similar values for the different (HSB) values. Under regular wave conditions, the mean (beta ) value for these four cases is 0.76, and under random wave conditions, the value reaches 0.68. This may indicate that the damping coefficient has reached its maximum value and no longer increases with increasing (HSB) value. To analyze this aspect in more detail, the wave height evolution measured for the four tests in which (beta ) reaches its maximum value are plotted (as shown in Supplementary Fig. S3). These tests correspond to Halimione sp. with a density of 100% and the shallowest water depth, h = 0.20 m. This species achieved the highest standing biomass value among the species considered in these experiments, and for h = 0.20 m, almost the entire water column was covered by vegetation. For these tests, a notable wave height attenuation was observed, where the wave height strongly decayed along the first 5 m of vegetation, and the wave height entirely dissipated along the last 4 m (as shown in Supplementary Fig. S3). The wave damping equation cannot suitably reproduce the strong wave decay within this few meters. Then, an almost constant wave damping coefficient value is reached under the different considered wave conditions, and a saturation regime is observed, in which the wave height beyond the meadow can be assumed to be negligible. To consider this phenomenon, a two-section fitting relationship is proposed, as shown in Fig. 6. The value of the saturation damping coefficient, chosen as the mean value of the four cases analyzed, is plotted as a dashed gray line, and a linear fit is obtained for the remaining data. The two-section fitting relationship is expressed in Eqs. (8) and (9) for both regular and random waves, respectively, where values between brackets are the 95% confidence interval for each coefficient.$$beta =left{begin{array}{ll}1.020 cdot {10}^{-3}left(1.112 cdot {10}^{-4}right)*HSB+0.088 ; (0.020) \ 0.758; (0.027)end{array}right. begin{array}{l} ;;0 < HSB < 659\ ;; HSB > 659end{array}$$
(8)
$$beta =left{begin{array}{l}1.310cdot {10}^{-3}left(1.232cdot {10}^{-4}right)*HSB+0.059; (0.017) \ 0.684 ;(0.066)end{array}right. begin{array}{l};;0474end{array}$$
(9)
All damping coefficients considered in the previous analysis were obtained without subtracting any additional source of dissipation such as bottom and wall friction. Previous works, such as24, highlighted the high importance of considering any other sources of wave dissipation besides the effect of vegetation elements when quantifying the wave height attenuation capacity. In this case, the flume walls were made of glass, and the friction induced by these walls could be considered negligible. However, bottom friction could be significant, as observed in tests run after removing all vegetation stems. Then, the wave damping coefficient obtained after subtracting the bottom friction contribution, ({beta }_{SB}), is studied here. Figure 7 shows the relationship obtained between this damping coefficient, ({beta }_{SB}), and hydraulic standing biomass, (HSB).Figure 7Wave damping coefficient, ({beta }_{SB}), as a function of the hydraulic standing biomass, (HSB), under all regular (top panel) and random (bottom panel) wave conditions. Both panels show linear fitting results obtained without considering the saturation point, indicated by the black solid line, and those obtained considering the saturation point, indicated by the gray solid line. The black dashed line indicates the saturation point. 95% confidence interval is shown in brackets and correlation coefficient (({rho }^{2})) for each fit is also displayed.Full size imageA linear relationship was also obtained for ({beta }_{SB}), revealing correlation coefficients similar to those obtained when analyzing (beta ). The obtained linear relationships under regular and random wave conditions are expressed as Eqs. (10) and (11), respectively, where values between brackets are the 95% confidence interval for each coefficient. A two-section fitting relationship, Eqs. (12) and (13), was also included considering the saturation regime obtained in the Halimione sp. 100% density and h = 0.20 m cases with a ({beta }_{SB}=) 0.69 and 0.63 under regular and random wave conditions, respectively.$${beta }_{SB}=1.051*{10}^{-3} left(7.063cdot {10}^{-5}right)*HSB$$
(10)
$${beta }_{SB}=1.296*{10}^{-3} left(6.894cdot {10}^{-5}right)*HSB$$
(11)
$${beta }_{SB}=left{begin{array}{l}1.151cdot {10}^{-3} left(7.445cdot {10}^{-5}right)*HSB \ 0.685 ;(0.047)end{array}right. begin{array}{l} ;; 0599end{array}$$
(12)
$${beta }_{SB}=left{begin{array}{l}1.396cdot {10}^{-3}left(7.919cdot {10}^{-5}right)*HSB \ 0.631 ;left(0.055right)end{array}right. begin{array}{l};; 0451end{array}$$
(13)
As can be noted, the ({beta }_{SB}) values are significantly lower than those obtained for (beta ), especially in the shallowest water depth cases where bottom friction is the highest, as discussed above. The estimation of (beta ) and ({beta }_{SB}) allows two possible approaches to determine the wave damping effect of a saltmarsh. The first approach, based on (beta ), includes wave damping induced by the combined effect of vegetation and bottom friction. Therefore, the consideration of (beta ) in analytical or numerical analysis could provide the total dissipation induced by the species under study, and sediment characteristics are not necessary for analysis. Considering that saltmarsh species grow in muddy to sandy environments and that the major contribution to the obtained wave attenuation is associated with vegetation, this approach may be the best option if soil properties are not thoroughly characterized.The second approach relies on the definition of ({beta }_{SB}). In this case, the wave damping contributions of vegetation drag and bottom friction are separated. Then, ({beta }_{SB}) can be used in cases where the effect of both momentum sinks can be separately evaluated. To quantify the wave damping contribution of vegetation drag only, ({beta }_{SB}) can be used, and then, the additional friction due to the bottom effect can be added considering the soil properties in each case. This second approach assumes a linear sum of both momentum sinks and could be applicable when soil properties are thoroughly characterized. More