Ecology

In the work reported here, we have examined the interaction of symbiotic partners representative of the three major diversification centers. Although P. vulgaris could establish symbiosis with diverse rhizobial lineages, Rhizobium etli seemed to predominate in nature in the bean nodules collected from the Americas8,9, while the Americas is where the origin and diversification of the host have been experimentally supported19,20. Genotypes other than R. etli that also induce nodule formation in the bean have already been taxonomically defined as species, for instance Rhizobium tropici and Rhizobium ecuadorense, both of which were isolated from areas in northwestern South America, namely Ecuador, Brazil, and Colombia.American-bean rhizobia, from soil samples retrieved by the common bean as well as isolates from nodules found in nature have possessed polymorphism in the nodC gene, disclosing three nodC genotypes namely α, (upgamma), and (updelta)9. The different nodC alleles in American strains exhibit a varying predominance in their regional distributions in correlation with the centers of bean genetic diversification. The nodC types α and (upgamma) were detected both in bean nodules and in soils from Mexico, whereas the nodC type (updelta) was clearly predominant in soil and nodules from the Southern Andes (i. e., in Bolivia and northwest Argentina9). A quantitatively balanced representation of rhizobia with nodC type α and (upgamma) was detected in soils from Ecuador, but the nodC type (upgamma) had been found to be predominantly isolated from nodules formed in nature in that area5,9,10. It should be noted that we have reported finding of equal distribution of allele nodC type α and γ among the nine R etli isolates from bean in Mexico reported by Silva et al.7,9. The occurrence of this polymorphism proved to contribute to examining rhizobial populations inhabiting the Americas and characterizing the interaction with beans in BGD centers from Mexico to the northwest of Argentina. In performing our nodC analysis, we were aware that rhizobia genes for symbiosis are carried on plasmids which might mediate horizontal transfer, however in agreement with Silva et al.7 we assumed that although genetic exchange could be important, it is not so extensive to prevent epidemic clones from arising at significant frequency. Similar findings were found in R. leguminosarum bv trifolii associated with native Trifolium species growing in nature21.Investigations in the last decade have proposed an evolutionary pathway for the host bean that provided us with a framework for examining our results on rhizobia-bean interactions and facilitated an interpretation of the results. The current model proposes the occurrence of a Mesoamerican origin from where dispersion by independent migrations over time led to the Mesoamerican and Andean gene pools and to the Ecuador-Peru wild common-bean populations2,19,20. We found a balanced competition between α and (upgamma) nodC types in beans from Mesoamerica and the southern Andes, whereas the beans from Ecuador and Peru revealed a clear affinity for nodulation with strains of nodC type α rather than with the sympatric strains nodC type (upgamma) that we assayed (R. ecuadorense, CIAT894 and Bra-5). Nevertheless, we have previously reported that native strains and isolates with respectively both nodC types α and (upgamma) were found in soils and bean nodules from Mexico9, whereas lineages harboring nodC type (upgamma) were found to be predominant in beans from the northern and central regions of Ecuador-Peru8,9. The present results, however, indicated a clear affinity of the Ecuadorean-Peruvian—i. e., AHD—beans for strains nodC type α when assessed for competition against nodC type (upgamma) (Fig. 2A). We also found that nodC type (updelta) displayed a clear predominant occupancy of nodules of the AHD beans in contrast to the scarce occupancy of nodules of the Mesoamerican and Andean beans (Fig. 2B). Taken together, these results indicate no affinity of AHD beans for sympatric rhizobial strains containing nodC type (upgamma)—despite the finding that rhizobia of nodC type (upgamma) appear to predominate in isolates of nodules formed in Ecuador9,10.We conclude that although rhizobial type nodC (upgamma) was previously found to predominate in bean nodules from Ecuador, the competitiveness of that rhizobial strain for nodulation compared to other genotypes of bean rhizobia was relatively low. A possible explanation could be that seeds might be assumed to play a key role as carriers during the dissemination of the bean throughout the American regions. Thus, we can hypothesize that at the time of bean dissemination both R. etli nodC types α and (upgamma) (R. ecuadorense and other lineages) moved in conjunction with the host from Mesoamerica to northern Ecuador-Peru, but the strains bearing nodC type (upgamma) achieved an adaptation—probably due to edaphic characteristics, environmental stresses, and other features—in such a way that that strain predominated in soils and succeeded in nodulation.Alternatively, that prevalence might arise from a host selection for a rhizobium that is more effective in nitrogen fixation in a new and different environment. A poor relationship, however, between competitiveness and efficiency was found in the pea22. In line with the concept of adaptation, the bean had been found to be preferentially nodulated by species of R. tropici in acidic soils in regions of Brazil and Africa4,23. Since the environment could also be a major influence on the host and its symbiotic interactions, the Andean area represents a cooler climate for the growth of the bean than the Mesoamerican region24,25. Furthermore, since our assays were performed in laboratory environment parameters, we do not rule out the effect -if any- by the diverse and complex soil microbial community coexisting with bean rhizobia. Within this context, two contrasting soils from Argentina which differ in geolocation and edaphic properties and the perlite substrate were assayed side by side in nodule occupancy of Negro Xamapa after inoculation with a mixture of strains nodC type α and γ (Results not shown). Our results showed that the predominance of nodC type γ in the occupation of the nodules of this variety (about 80% occupation) is not affected by the type of substrate (p = 0.5566). Yet, we assume that the performance in diverse soil and ecosystems should be further evaluated in situ. In agreement, a good coevolution of rhizobia strains with nodC type (upgamma) was detected in nodules of bean varieties from the Mesoamerica and Andean genetic pools inoculated with soil samples from Mexico, Ecuador, and Northwest of Argentina, respectively (see Table 2 in Aguilar et al., 2004) [9].With respect to the interaction in the southern Andes, we propose another interpretation that takes into consideration the bottleneck that occurred before domestication in the Andes, as was indicated by Bitocchi et al.26, which scenario enables the assumption that the adaptation and concomitant diversification involved a coevolution of the symbioses. Therefore, similar profiles of competitiveness for nodulation in Mesoamerican and Andean beans were found between nodC type (upgamma) versus nodC types α and (updelta), but a significant occupancy by the nodC type (updelta) was recorded in the Andean beans.Our work suggests that the genetics of both the host and the bacteria determine the mutual affinity and additionally indicates that symbiotic interaction is another trait of legumes sensitive to the effects of evolution and ecological adaptation to the locale environment such as the characteristics of the soil and the climate.The analysis of the genetic sequences of the bean that encode genes associated with symbiosis, revealed variation of NFR1, NFR5 and NIN over the representative accessions of the Mesoamerican, the Andean, and the AHD gene pools. It is proposed that a receptor complex composed of NFR1 and NFR5 initiates signal transduction in response to Nod-factor synthesized and released by rhizobia27. Although the variation consisted mainly in neutral-amino-acid substitutions, thus suggesting only minimal changes in the functionality, if any at all; we could cite the convincing and relevant evidence reported by Radutoiu et al.27 that the amino-acid residue 118 of the second LysM module of NFR5 is essential for the recognition of rhizobia by species of Lotus japonicus and Lotus filicaulis. Our finding that the Mesoamerican-bean NFR5 has glutamine (Q) in position 151, whereas the Andean and the AHD both have proline (P)—neither of which amino acids is neutral—would merit further investigation to evaluate if such a mutation might play a role in nodulation preference. Although this result must be considered with caution, we found that the conserved polymorphism in the NFR1 and NFR5 proteins has caused the beans representative of the genetic pool Ecuador-Peru—i. e., the AHD—to be grouped in a cluster separate from those of Mesoamerica and the Andes. What we found to be interesting was that the phylogenetic and RMSD profiles of grouping the sequences are consistent with different evolutionary pathways in beans from the AHD and the Andean areas. This observation agrees with the proposal of Randón-Anaya et al.2 that those former beans from northern Peru-Ecuador originated from an ancestral form earlier than that of Mesoamerican- and Andean-bean genotypes. In addition, by applying a suppressive subtractive hybridization approach a set of bean genes were identified in our laboratory to be expressed in early step of infection by the cognate rhizobia28. Taken these results together, we conclude that genomic regions and patterns of expression in the host appear associated with an affinity for nodulation.Within a broader context, we believe that our results on the biogeography of bean-rhizobia interactions in the region where the origin and domestication of the host plants occurred provide novel useful issues to be considered in inoculation programs, for instance those involving selection of strains and cultivars, and invite to validate these findings in follow up field trials. More

Case, T. J., Bolger, D. T. & Richman, A. D. Reptilian extinctions: The last ten thousand years. In Conservation Biology (eds Fiedler, P. L. & Jain, S. K.) 91–125 (Springer, 1992).
Google Scholar 
Shivanna, K. R. The sixth mass extinction crisis and its impact on biodiversity and human welfare. Resonance 25, 93–109 (2020).
Google Scholar 
Ceballos, G., Ehrlich, P. R., Barnosky, A. D., García, A., Pringle, R. M. & Palmer, T. M. Accelerated modern human–induced species losses: Entering the sixth mass extinction. Sci. Adv. 1, (2015)Lawler, J. J. et al. Conservation science: A 20-year report card. Front. Ecol. Environ. 4, 473–480 (2006).
Google Scholar 
Sodhi, N. S., Brook, B. W. & Bradshaw, C. J. A. Tropical Conservation Biology (Wiley-Blackwell, 2007).
Google Scholar 
Scheffers, B. R., Yong, D. L., Harris, J. B. C., Giam, X. & Sodhi, N. S. The world’s rediscovered species: Back from the brink?. PLoS ONE 6, 1–8 (2011).
Google Scholar 
Ineich I. Bocourt’s terrific skink, Phoboscincus bocourti Brocchi, 1876 (Squamata, Scincidae, Lygosominae). In 7. Biodiversity studies in New Caledonia.Mémoires du Muséum National d’Histoire Naturelle (ed. Grancolas, P.) vol. 198, 149–174, Muséum National d’Histoire Naturelle, (2009).Holden, M. & Ineich, I. scinque terrifiant terrifié. Le Courrier de la Nat. 312, 4 (2018).
Google Scholar 
Sadlier, R. A., Deuss, M., Bauer, A. M. & Jourdan, H. Kuniesaurus albiauris, a new genus and species of scincid lizard from the Île des Pins, New Caledonia, with comments on the diversity and affinities of the region’s lizard fauna. Pac. Sci. 73, 123–141 (2019).Bauer, A. M. & Sadlier, R. A. Lizard discoveries and rediscoveries in the New Caledonian region. In Flores, O., Ah-Peng, C., & Wilding, N. Island Biology 2019. Third International Conference on Island Ecology, Evolution and Conservation: Book of Abstracts. Island Biology 2019, Jul 2019, Saint Denis, France. 2020. ffhal-02633975v2 243 (2019).Ineich, I., Sadlier, R. A., Bauer, A. M., Jackman, T. R. & Smith, S. A. Bocourt’s terrific skink, Phoboscincus bocourti (Brocchi, 1876), and the monophyly of the genus Phoboscincus Greer, 1974. In Zoologia Neocaledonica 8. Biodiversity studies in New Caledonia. Mémoires du Muséum National d’Histoire Naturelle (eds Guilbert, E. et al.) 69–78 Muséum National d’Histoire naturelle, (2014).
Google Scholar 
Caut, S., Holden, M., Jowers, M. J., Boistel, R. & Ineich, I. Is Bocourt’s terrific skink really so terrific? Trophic myth and reality. PLoS One 8, e78638 (2013).Sagonas, K. et al. Insularity affects head morphology, bite force and diet in a Mediterranean lizard. Biol. J. Linn. Soc. 112, 469–484 (2014).
Google Scholar 
Tseng, W.-H. et al. Opsin gene expression regulated by testosterone level in a sexually dimorphic lizard. Sci. Rep. 8, 16055 (2018).ADS 
PubMed 
PubMed Central 

Google Scholar 
Avramo, V. et al. Evaluating the island effect on phenotypic evolution in the Italian wall lizard, Podarcis siculus (Reptilia: Lacertidae). Biol. J. Linn. Soc. 132, 655–665 (2021).
Google Scholar 
Siliceo-Cantero, H. H., Benítez-Malvido, J. & Suazo-Ortuño, I. Insularity effects on the morphological space and sexual dimorphism of a tropical tree lizard in western Mexico. J. Zool. 311, 277–285 (2020).
Google Scholar 
Pérez-Mellado, V. & Corti, C. Dietary adaptations and herbivory in lacertid lizards of the genus Podarcis from western Mediterranean islands (Reptilia: Sauria). Bonner Zool. Beiträge 44, 193–220 (1993).
Google Scholar 
Castilla, A. M., Vanhooydonck, B. & Catenazzi, A. Feeding behavior of the Columbretes lizard Podarcis atrata, in relation to the marine species, Ligia italica (Isopoda, Crustaceae). Belgian J. Zool. 138, 146–148 (2008).
Google Scholar 
Castilla, A. M. & Herrel, A. The scorpion Buthus occitanus as a profitable prey for the endemic lizard Podarcis atrata in the volcanic Columbretes islands (Mediterranean, Spain). J. Arid Environ. 73, 378–380 (2009).ADS 

Google Scholar 
Van Damme, R. Evolution of herbivory in lacertid lizards: Effects of insularity and body size. J. Herpetol. 33, 663 (1999).
Google Scholar 
Pafilis, P., Meiri, S., Foufopoulos, J. & Valakos, E. Intraspecific competition and high food availability are associated with insular gigantism in a lizard. Naturwissenschaften 96, 1107–1113 (2009).ADS 
CAS 
PubMed 

Google Scholar 
D’Amore, D. C. et al. Increasing dietary breadth through allometry: Bite forces in sympatric Australian skinks. Herpetol. Notes 11, 179–187 (2018).
Google Scholar 
Taverne, M. et al. Proximate and ultimate drivers of variation in bite force in the insular lizards Podarcis melisellensis and Podarcis sicula. Biol. J. Linn. Soc. 131, 88–108 (2020).
Google Scholar 
Kingsolver, J. G. & Pfennig, D. W. Patterns and power of phenotypic selection in nature. Bioscience 57, 561–572 (2007).
Google Scholar 
Itescu, Y., Foufopoulos, J., Pafilis, P. & Meiri, S. The diverse nature of island isolation and its effect on land bridge insular faunas. Glob. Ecol. Biogeogr. 29, 262–280 (2020).
Google Scholar 
Polis, G. A. & Hurd, S. D. Linking marine and terrestrial food webs: Allochthonous input from the ocean supports high secondary productivity on small islands and coastal land communities. Am. Nat. 147, 396–423 (1996).
Google Scholar 
Donihue, C. M., Brock, K. M., Foufopoulos, J. & Herrel, A. Feed or fight: Testing the impact of food availability and intraspecific aggression on the functional ecology of an island lizard. Funct. Ecol. 30, 566–575 (2016).
Google Scholar 
Runemark, A., Sagonas, K. & Svensson, E. I. Ecological explanations to island gigantism: Dietary niche divergence, predation, and size in an endemic lizard. Ecology 96, 2077–2092 (2015).PubMed 

Google Scholar 
Verwaijen, D., Van Damme, R. & Herrel, A. Relationships between head size, bite force, prey handling efficiency and diet in two sympatric lacertid lizards. Funct. Ecol. 16, 842–850 (2002).
Google Scholar 
Herrel, A., O’Reilly, J. C. & Richmond, A. M. Evolution of bite performance in turtles. J. Evol. Biol. 15, 1083–1094 (2002).
Google Scholar 
Herrel, A., Vanhooydonck, B., Joachim, R. & Irschick, D. J. Frugivory in polychrotid lizards: Effects of body size. Oecologia 140, 160–168 (2004).ADS 
CAS 
PubMed 

Google Scholar 
Herrel, A., Vanhooydonck, B. & Van Damme, R. Omnivory in lacertid lizards: Adaptive evolution or constraint?. J. Evol. Biol. 17, 974–984 (2004).CAS 
PubMed 

Google Scholar 
Herrel, A., Podos, J., Huber, S. K. & Hendry, A. P. Bite performance and morphology in a population of Darwin’s finches: Implications for the evolution of beak shape. Funct. Ecol. 19, 43–48 (2005).
Google Scholar 
Herrel, A., Podos, J., Huber, S. K. & Hendry, A. P. Evolution of bite force in Darwin’s finches: A key role for head width. J. Evol. Biol. 18, 669–675 (2005).CAS 
PubMed 

Google Scholar 
Aguirre, L. F., Herrel, A., Van Damme, R. & MatThysen, E. The implications of food hardness for diet in bats. Funct. Ecol. 17, 201–212 (2003).
Google Scholar 
Herrel, A. & Holanova, V. Cranial morphology and bite force in Chamaeleolis lizards—Adaptations to molluscivory?. Zoology 111, 467–475 (2008).PubMed 

Google Scholar 
Greer, A. E. Distribution of maximum snout-vent length among species of scincid lizards. J. Herpetol. 35, 383 (2001).
Google Scholar 
Burggren, W. W. & McMahon, B. R. Biology of the Land Crabs, Cambridge University Press, (1988).
Google Scholar 
Grubb, P. Ecology of terrestrial decapod crustaceans on Aldabra. Philos. Trans. R. Soc. Lond. B Biol. Sci. 260, 411–416 (1971)Wineski, L. E. & Gans, C. Morphological basis of the feeding mechanics in the shingle-back lizard Trachydosaurus rugosus (Scincidae, Reptilia). J. Morphol. 181, 271–295 (1984).CAS 
PubMed 

Google Scholar 
Herrel, A., Verstappen, M. & De Vree, F. Modulatory complexity of the feeding repertoire in scincid lizards. J. Comp. Physiol. A Sens. Neural Behav. Physiol. 184, 501–518 (1999).Herrel, A., Aerts, P. & De Vree, F. Ecomorphology of the lizard feeding apparatus: A modelling approach. Neth. J. Zool. 48, 1–25 (1998).
Google Scholar 
Hartnoll, R. G. Evolution, systematics, and geographical distribution. In Biology of the Land Crabs (eds Burggren, W. W. & McMahon, B. R.) 6–54, (Cambridge University Press, 1988).
Google Scholar 
Ben-David, M. & Schell, D. M. Mixing models in analyses of diet using multiple stable isotopes: A response. Oecologia 127, 180–184 (2001).ADS 
CAS 
PubMed 

Google Scholar 
Caut, S., Angulo, E. & Courchamp, F. Caution on isotopic model use for analyses of consumer diet. Can. J. Zool. 86, 438–445 (2008).CAS 

Google Scholar 
Warne, R. W., Gilman, C. A. & Wolf, B. O. Tissue-carbon incorporation rates in lizards: Implications for ecological studies using stable isotopes in terrestrial ectotherms. Physiol. Biochem. Zool. 83, 608–617 (2010).PubMed 

Google Scholar 
Steinitz, R., Lemm, J. M., Pasachnik, S. A. & Kurle, C. M. Diet-tissue stable isotope (Δ13C and Δ15N) discrimination factors for multiple tissues from terrestrial reptiles. Rapid Commun. Mass Spectrom. 30, 9–21 (2016).ADS 
CAS 
PubMed 

Google Scholar 
Lattanzio, M. & Miles, D. Stable carbon and nitrogen isotope discrimination and turnover in a small-bodied insectivorous lizard. Isot. Environ. Health Stud. 52, 673–681 (2016).CAS 

Google Scholar 
Durso, A. M., Smith, G. D., Hudson, S. B. & French, S. S. Stoichiometric and stable isotope ratios of wild lizards in an urban landscape vary with reproduction, physiology, space and time. Conserv. Physiol. 8, 1–14 (2020).
Google Scholar 
Warne, R. W. & Wolf, B. O. Nitrogen stable isotope turnover and discrimination in lizards. Rapid Commun. Mass Spectrom. 35, e9030 (2021).Aerts, P., De Vree, F. & Herrel, A. Ecomorphology of the lizard feeding apparatus: A modelling approach. Neth. J. Zool. 48, 1–25 (1997).
Google Scholar 
Herrel, A., Schaerlaeken, V., Meyers, J. J., Metzger, K. A. & Ross, C. F. The evolution of cranial design and performance in squamates: Consequences of skull-bone reduction on feeding behavior. Integr. Comp. Biol. 47, 107–117 (2007).PubMed 

Google Scholar 
Beuttner, A. & Koch, C. Analysis of diet composition and morphological characters of the Peruvian lizard Microlophus stolzmanni (Squamata: Tropiduridae). Phyllomedusa J. Herpetol. 18, 47–62 (2019).
Google Scholar 
Herrel, A., Aerts, P. & Vree, D. Static biting in lizards: Functional morphology of the temporal ligaments. J. Zool. 244, 135–143 (1998).
Google Scholar 
Greer, A. The genetic relationships of the scincid lizard genus Leiolopisma and its relatives. Aust. J. Zool. Suppl. Ser. 22, 1–67 (1974).
Google Scholar 
Shirley, M. H., Carr, A. N., Nestler, J. H., Vliet, K. A. & Brochu, C. A. Systematic revision of the living African slender-snouted crocodiles (Mecistops Gray, 1844). Zootaxa 4504, 151 (2018).PubMed 

Google Scholar 
Yoshioka, S. & Kimura, T. What does the red-eared slider eat on the tidal flats? Comparing the diet of the invasive alien species Trachemys scripta elegans inhabiting the tidal flat and freshwaters. Jpn. J. Benthol. 72, 83–93 (2018).
Google Scholar 
Bernal, S. & Magda, S. Análisis de los contenidos estomacales de las tortugas y cachirres de la Reserva Natural Privada de la Sociedad Civil Bojonawi (Puerto Carreño, Vichada). (Bogotá: Instituto de Investigación de Recursos Biológicos Alexander von Humboldt, 2020).Murphy, J. C. Homalopsid Snakes, Evolution in the Mud (Krieger Publishing Company, 2007).
Google Scholar 
Chen, P. Z. An observation of crab predation by a Gerard’s water snake, Gerarda prevostiana (Reptilia: Squamata: Homalopsidae) in the wild at Sungei Buloh, Singapore. Nat. Singap. 3, 195–197 (2010).
Google Scholar 
Jayne, B. C., Voris, H. K. & Ng, P. K. L. Snake circumvents constraints on prey size. Nature 418, 143–143 (2002).ADS 
CAS 
PubMed 

Google Scholar 
Jayne, B. C., Voris, H. K. & Ng, P. K. L. How big is too big? Using crustacean-eating snakes (Homalopsidae) to test how anatomy and behaviour affect prey size and feeding performance. Biol. J. Linn. Soc. 123, 636–650 (2018).
Google Scholar 
Murphy, J. C. & Voris, H. K. Aquatic snakes with crustacean-eating habits elude herpetologists for two centuries. Litt. Serpentium 22, 107–114 (2002).
Google Scholar 
Voris, H. K. & Murphy, J. C. The prey and predators of Homalopsine snakes. J. Nat. Hist. 36, 1621–1632 (2002).
Google Scholar 
Wai-Neng, L. & Melville, D. S. Notes on the feeding of Enhydris bennetti (Gray) (Reptilia, Squamata, Colubridae) in Hong Kong. Mem. Hong Kong Nat. Hist. Soc. 19, 117 (2020).
Google Scholar 
López-Hurtado, Y., García-Padrón, L. Y., González, A., Díaz, L. M. & Rodríguez-Cabrera, T. M. Notes on the feeding habits of the Caribbean watersnake, Tretanorhinus variabilis (Dipsadidae). Reptil. Amphib. 27, 147–153 (2020).
Google Scholar 
Gripshover, N. D. & Jayne, B. C. Crayfish eating in snakes: Testing how anatomy and behavior affect prey size and feeding performance. Integr. Org. Biol. 3, 1–16 (2021).
Google Scholar 
Naish, D. The Madagascan skink Amphiglossus eats crabs. Sci. Am. Blog Netw. https://blogs.scientificamerican.com/tetrapod-zoology/the-madagascan-skink-amphiglossus-eats-crabs/ (2016).Hediger, H. Beitrag zur herpetologie und zoogeographie Neu Britanniens und einiger umliegender gebiete. Zool. Jahrbücher. Abteilung für Syst. Geogr. und Biol. der Tiere 65, 441–582 (1934).McCoy, M. W. Reptiles of the Solomon Islands, (Pensoft Publishers, 2006).
Google Scholar 
Huang, W. S. Ecology and reproductive patterns of the littoral skink Emoia atrocostata on an East Asian tropical rainforest island. Zool. Stud. 50, 506–512 (2011).
Google Scholar 
Anderson, C. Decapod crustacean species of Aride Island, Seychelles. Phelsuma 2(12), 36–49 (1994).
Google Scholar 
Paulay, G. & Starmer, J. Evolution, insular restriction, and extinction of oceanic land crabs, exemplified by the loss of an endemic Geograpsus in the Hawaiian Islands. PLoS ONE 6, e19916 (2011).Cleuren, J., Aerts, P. & de Vree, F. Bite and joint force analysis in Caiman crocodilus. Belgian J. Zool. 125, 79–94 (1995).
Google Scholar 
Meyers, J. J., Nishikawa, K. C. & Herrel, A. The evolution of bite force in horned lizards: The influence of dietary specialization. J. Anat. 232, 214–226 (2018).PubMed 

Google Scholar 
Van Damme, R., De Vree, F. & Herrel, A. Sexual dimorphism of head size in Podarcis hispanica atrata: Testing the dietary divergence hypothesis by bite force analysis. Neth. J. Zool. 46, 253–262 (1995).
Google Scholar 
Gröning, F. et al. The importance of accurate muscle modelling for biomechanical analyses: A case study with a lizard skull. J. R. Soc. Interface 10, 1–10 (2013).
Google Scholar 
Vanhooydonck, B., Boistel, R., Fernandez, V. & Herrel, A. Push and bite: Trade-offs between burrowing and biting in a burrowing skink (Acontias percivali). Biol. J. Linn. Soc. 102, 91–99 (2011).
Google Scholar 
Handschuh, S. et al. Cranial kinesis in the miniaturised lizard Ablepharus kitaibelii (Squamata: Scincidae). J. Exp. Biol. 222, 1–15 (2019).
Google Scholar 
Le Guilloux, M. et al. Trade-offs between burrowing and biting force in fossorial scincid lizards?. Biol. J. Linn. Soc. 130, 310–319 (2020).
Google Scholar 
Herrel, A, Spithoven, L., Van Damme, R. & De Vree, F. Sexual dimorphism of head size in Gallotia galloti: Testing the niche divergence hypothesis by functional analyses. Funct. Ecol. 13, 289–297 (1999).
Google Scholar 
Herrel, A., De Grauw, E. & Lemos-Espinal, J. A. Head shape and bite performance in xenosaurid lizards. J. Exp. Zool. 290, 101–107 (2001).CAS 
PubMed 

Google Scholar 
Herrel, A., Petrochic, S. & Draud, M. Sexual dimorphism, bite force and diet in the diamondback terrapin. J. Zool. 304, 217–224 (2018).
Google Scholar  More

Stochastic age-specific transmission modelWe formulate a stochastic age-specific transmission model in the general Susceptible(S)-Exposed(E)-Reported(I)-Unreported(U)-Recovered(R) framework. For a particular age group (i) at time (t-1) ((i=1) corresponding to the 0–17 years, (i=2) to 18–44, (i=3) to 45–64 and (i=4) to 65+), we have$$begin{array}{l}{S}_{i}(t)= {S}_{i}(t-1)-{n}_{S{E}_{i}}(t)\ {E}_{i}(t)= {E}_{i}(t-1)+{n}_{S{E}_{i}}(t)-\ {n}_{E{I}_{i}}(t)-{n}_{E{U}_{i}}(t)\ {I}_{i}(t)= {I}_{i}(t-1)+{n}_{E{I}_{i}}(t)-{n}_{I{R}_{i}}(t)\ {U}_{i}(t)= {U}_{i}(t-1)+{n}_{E{U}_{i}}(t)-{n}_{U{R}_{i}}(t)\ {R}_{i}(t)= {R}_{i}(t-1)+{n}_{I{R}_{i}}(t)+{n}_{U{R}_{i}}(t),end{array}$$
(1)
where ({n}_{{XY}_{i}}(t)) represents number of transitions between a class X and class Y for age group (i) at time (t).The number of transitions from the susceptible to exposed class for group (i) at time (t) is modelled by$$begin{aligned}{n}_{S{E}_{i}}(t)&sim Poi({S}_{i}(t-1)times {gamma }_{i}(t)times \ & quad sum_{j=1}beta (t)times {c}_{j,i}(t)times {{I}_{j}(t-1)+{U}_{j}(t-1)}).end{aligned}$$
(2)
Here, (beta (t)) denotes the average infectiousness of an infectious individual and ({c}_{j,i}(t)) is the average number of contacts per day made by age group (j) to (i). Also note that the product (beta (t)times {c}_{j,i}(t)) may represent age-specific transmissibility (of age group (j)) accounting for contacts. We allow and infer two change points of (beta (t)) (one potentially correlates to changes due to the implementation of lockdown and another one to changes due to the lifting of lockdown), i.e.,$$beta left(tright)=left{begin{array}{ll}{beta }_{0},&quad if; tle {T}_{1}\ {beta }_{1}={omega }_{1}times {beta }_{0},&quad if ;{T}_{1}{T}_{2},end{array}right.$$
(3)
where ({T}_{1}) and ({T}_{2}) are the two change points to be inferred (({T}_{2}ge {T}_{1})). ({gamma }_{i}(t)) denotes the susceptibility of group (i) relative to the oldest age group (i.e., ({gamma }_{4}=1)), which is also allowed to change proportionally after lifting the lockdown. Note that ({gamma }_{i}(t)) implicitly incorporates any behavioral effects (e.g., potential reduction of risk of getting infection due to facemask wearing). Transitions between other classes are modelled as:$$begin{aligned}{n}_{E{U}_{i}}(t)sim & Bin({n}_{S{E}_{i}}(t-{D}_{EU}),{p}_{{U}_{i}}(t-{D}_{EU}))\ {n}_{E{I}_{i}}(t)=& {n}_{S{E}_{i}}(t-{D}_{EI})-{n}_{E{U}_{i}}(t)\ {n}_{I{R}_{i}}(t)=& {n}_{E{I}_{i}}(t-{D}_{IR})\ {n}_{U{R}_{i}}(t)=& {n}_{E{U}_{i}}(t-{D}_{UR}),end{aligned}$$
(4)
where ({D}_{EI}), ({D}_{EU}), ({D}_{IR}) and ({D}_{UR}) denote the mean waiting times between the indicated two classes. We assume that ({D}_{EI})= ({D}_{EU})=7 days and ({D}_{IR})= ({D}_{UR})=14 days. ({p}_{{U}_{i}}(t)) represents probability that an infection is unreported at times (t) for age group (i), we assume$${p}_{{U}_{i}}(t)=1-frac{{e}^{{f}_{i}(t)}}{1+{e}^{{f}_{i}(t)}}.$$
(5)
({f}_{i}(.)) is an increasing function with ({f}_{i}(t)={a}_{i}+{b}_{i}times t), where (-infty More

Analysing the relations between FTSE100 and self-reported measures of emotional well-being we confirmed that market ups (higher FTSE100 scores) were associated with higher scores of “happiness” and lower scores in self-reported “negative emotional facets”: irritability, hurt and nervous feelings, anxiety (Fig. 1; Table 1). The identified association also held true for the 5.5-years of the MRI subsample (Supplementary Table S2). We further explored non-imaging variables that are associated with mood changes, i.e. alcohol intake (overall intake frequency and a composite score reflecting weekly intake of all alcoholic beverages) and diastolic blood pressure (automatic readings in mmHg measured at rest), and showed that they were also highly correlated with the FTSE100 (Fig. 1A) in that both measures increased when the stock market decreased in value. Several of these effects (relation between stock market and negative emotions, blood pressure or alcohol-intake) were reproduced in the My Connectome data-set consisting of one single subject whose measurements were taken at 81 timepoints during a period or 1.5 years (Fig. 1B).Figure 1Non-MRI variables and stock market moves. The figure illustrates the identified associations between stock market moves and non-MRI indicators of well-being in the UK Biobank sample (top panel A) and My Connectome data, a single-subject study (bottom panel B); *p  More

De-extinction efforts that use genome editing aim to identify the genome sequence of extinct species and then edit the genome of a closely related, living species. Lin et al. explored the feasibility of this approach by sequencing ancient DNA samples of the extinct Christmas Island rat (Rattus macleari), which had been originally collected between 1900–1902. The authors then mapped the resulting sequence to reference genomes of different living Rattus species. Even when using the high-quality Norway brown rat (Rattus norvegicus) as a reference, the team found that nearly 5% of the genome sequence was unmappable owing to evolutionary divergence of the two species. Of note, the incompletely covered genomic regions were not random but disproportionately affected immune response and olfaction genes, which would have implications for the biology of any reconstructed animals. More

Okogwu, O. I., Nwonumara, G. N. & Okoh, F. A. Evaluating heavy metals pollution and exposure risk through the consumption of four commercially important fish species and water from cross river ecosystem, Nigeria. Bull. Environ. Contam. Toxicol. 102, 867–872. https://doi.org/10.1007/s00128-019-02610-4 (2019).CAS 
Article 
PubMed 

Google Scholar 
Fuentes-Gandara, F., Pinedo-Hernández, J., Marrugo-Negrete, J. & Díez, S. Human health impacts of exposure to metals through extreme consumption of fish from the Colombian Caribbean Sea. Environ. Geochem. Health 40, 229–242. https://doi.org/10.1007/s10653-016-9896-z (2018).CAS 
Article 
PubMed 

Google Scholar 
Liu, X. et al. Human health risk assessment of heavy metals in soil–vegetable system: A multi-medium analysis. Sci. Total Environ. 463, 530–540 (2013).PubMed 

Google Scholar 
Lelieveld, J., Evans, J. S., Fnais, M., Giannadaki, D. & Pozzer, A. The contribution of outdoor air pollution sources to premature mortality on a global scale. Nature 525, 367–371. https://doi.org/10.1038/nature15371 (2015).CAS 
Article 
PubMed 

Google Scholar 
Huang, R.-J. et al. High secondary aerosol contribution to particulate pollution during haze events in China. Nature 514, 218–222. https://doi.org/10.1038/nature13774 (2014).CAS 
Article 
PubMed 

Google Scholar 
Rajeshkumar, S. et al. Studies on seasonal pollution of heavy metals in water, sediment, fish and oyster from the Meiliang Bay of Taihu Lake in China. Chemosphere 191, 626–638. https://doi.org/10.1016/j.chemosphere.2017.10.078 (2018).CAS 
Article 
PubMed 

Google Scholar 
Gao, X. & Chen, C.-T.A. Heavy metal pollution status in surface sediments of the coastal Bohai Bay. Water Res. 46, 1901–1911. https://doi.org/10.1016/j.watres.2012.01.007 (2012).CAS 
Article 
PubMed 

Google Scholar 
Naser, H. A. Assessment and management of heavy metal pollution in the marine environment of the Arabian Gulf: A review. Mar. Pollut. Bull. 72, 6–13. https://doi.org/10.1016/j.marpolbul.2013.04.030 (2013).CAS 
Article 
PubMed 

Google Scholar 
Zhang, Y. et al. Heavy metals in aquatic organisms of different trophic levels and their potential human health risk in Bohai Bay, China. Environ. Sci. Pollut. Res. 23, 17801–17810 (2016).CAS 

Google Scholar 
Wei, M., Yanwen, Q., Zheng, B. & Zhang, L. Heavy metal pollution in Tianjin Bohai bay, China. J. Environ. Sci. 20, 814–819 (2008).
Google Scholar 
Zhao, B. et al. Spatiotemporal variation and potential risks of seven heavy metals in seawater, sediment, and seafood in Xiangshan Bay, China (2011–2016). Chemosphere 212, 1163–1171. https://doi.org/10.1016/j.chemosphere.2018.09.020 (2018).CAS 
Article 
PubMed 

Google Scholar 
Wang, Y. & Fang, X. Analysis of the impact of heavy metal on the Chinese aquaculture and the ecological hazard. GuangDong 836, 156.152 (2016).
Google Scholar 
Pini, J., Richir, J. & Watson, G. Metal bioavailability and bioaccumulation in the polychaete Nereis (Alitta) virens (Sars): The effects of site-specific sediment characteristics. Mar. Pollut. Bull. 95, 565–575 (2015).CAS 
PubMed 

Google Scholar 
Amoozadeh, E. et al. Marine organisms as heavy metal bioindicators in the Persian Gulf and the Gulf of Oman. Environ. Sci. Pollut. Res. 21, 2386–2395 (2014).CAS 

Google Scholar 
Gu, Y.-G., Huang, H.-H., Liu, Y., Gong, X.-Y. & Liao, X.-L. Non-metric multidimensional scaling and human risks of heavy metal concentrations in wild marine organisms from the Maowei Sea, the Beibu Gulf, South China Sea. Environ. Toxicol. Pharmacol. 59, 119–124. https://doi.org/10.1016/j.etap.2018.03.002 (2018).CAS 
Article 
PubMed 

Google Scholar 
Kennedy, A., Martinez, K., Chuang, C.-C., LaPoint, K. & McIntosh, M. Saturated fatty acid-mediated inflammation and insulin resistance in adipose tissue: Mechanisms of action and implications. J. Nutr. 139, 1–4. https://doi.org/10.3945/jn.108.098269 (2008).CAS 
Article 
PubMed 

Google Scholar 
Hao, Z. et al. Heavy metal distribution and bioaccumulation ability in marine organisms from coastal regions of Hainan and Zhoushan, China. Chemosphere 226, 340–350. https://doi.org/10.1016/j.chemosphere.2019.03.132 (2019).CAS 
Article 
PubMed 

Google Scholar 
Golden, C. D. et al. Nutrition: Fall in fish catch threatens human health. Nat. News 534, 317 (2016).
Google Scholar 
Bosch, A. C., O’Neill, B., Sigge, G. O., Kerwath, S. E. & Hoffman, L. C. Heavy metals in marine fish meat and consumer health: A review. J. Sci. Food Agric. 96, 32–48 (2016).CAS 
PubMed 

Google Scholar 
Burger, J., Gochfeld, M., Jeitner, C., Pittfield, T. & Donio, M. Heavy metals in fish from the Aleutians: Interspecific and locational differences. Environ. Res. 131, 119–130. https://doi.org/10.1016/j.envres.2014.02.016 (2014).CAS 
Article 
PubMed 

Google Scholar 
Anandkumar, A., Nagarajan, R., Prabakaran, K., Chua Han, B. & Rajaram, R. Human health risk assessment and bioaccumulation of trace metals in fish species collected from the Miri coast, Sarawak, Borneo. Mar. Pollut. Bull. 133, 655–663. https://doi.org/10.1016/j.marpolbul.2018.06.033 (2018).CAS 
Article 

Google Scholar 
Murtala, B. A., Abdul, W. O. & Akinyemi, A. A. Bioaccumulation of heavy metals in fish (Hydrocynus forskahlii, Hyperopisus bebe occidentalis and Clarias gariepinus) organs in downstream Ogun coastal water, Nigeria. J. Agric. Sci. 4, 51 (2012).
Google Scholar 
Ahmed, A. S. S., Rahman, M., Sultana, S., Babu, S. M. O. F. & Sarker, M. S. I. Bioaccumulation and heavy metal concentration in tissues of some commercial fishes from the Meghna River Estuary in Bangladesh and human health implications. Mar. Pollut. Bull. 145, 436–447. https://doi.org/10.1016/j.marpolbul.2019.06.035 (2019).CAS 
Article 
PubMed 

Google Scholar 
Sun, X. et al. Source identification, geochemical normalization and influence factors of heavy metals in Yangtze River Estuary sediment. Environ. Pollut. 241, 938–949. https://doi.org/10.1016/j.envpol.2018.05.050 (2018).CAS 
Article 
PubMed 

Google Scholar 
Dadar, M., Adel, M., NasrollahzadehSaravi, H. & Fakhri, Y. Trace element concentration and its risk assessment in common kilka (Clupeonella cultriventris caspia Bordin, 1904) from southern basin of Caspian Sea. Toxin Rev. 36, 222–227 (2017).CAS 

Google Scholar 
Chakraborty, P., Raghunadh Babu, P. V., Acharyya, T. & Bandyopadhyay, D. Stress and toxicity of biologically important transition metals (Co, Ni, Cu and Zn) on phytoplankton in a tropical freshwater system: An investigation with pigment analysis by HPLC. Chemosphere 80, 548–553. https://doi.org/10.1016/j.chemosphere.2010.04.039 (2010).CAS 
Article 
PubMed 

Google Scholar 
Handy, R. Seminar Series-Society for Experimental Biology 29–60 (Cambridge University Press, 1997).
Google Scholar 
Ahmed, M. K. et al. Human health risks from heavy metals in fish of Buriganga river, Bangladesh. Springerplus 5, 1–12 (2016).
Google Scholar 
WHO. Heavy metals-environmental aspects. Environment Health Criteria. No. 85. (1989).Xu, H. et al. Long-term study of heavy metal pollution in the northern Hangzhou Bay of China: Temporal and spatial distribution, contamination evaluation, and potential ecological risk. Environ. Sci. Pollut. Res. 28, 10718–10733 (2021).CAS 

Google Scholar 
El-Moselhy, K. M., Othman, A. I., AbdEl-Azem, H. & El-Metwally, M. E. A. Bioaccumulation of heavy metals in some tissues of fish in the Red Sea, Egypt. Egypti. J. Basic Appl. Sci. 1, 97–105. https://doi.org/10.1016/j.ejbas.2014.06.001 (2014).Article 

Google Scholar 
Jezierska, B. & Witeska, M. Soil and Water Pollution Monitoring, Protection and Remediation 107–114 (Springer, 2006).
Google Scholar 
Bawuro, A. A., Voegborlo, R. B. & Adimado, A. A. Bioaccumulation of heavy metals in some tissues of fish in Lake Geriyo, Adamawa State, Nigeria. J. Environ. Public Health 2018, 1854892. https://doi.org/10.1155/2018/1854892 (2018).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Zhuang, P., McBride, M. B., Xia, H., Li, N. & Li, Z. Health risk from heavy metals via consumption of food crops in the vicinity of Dabaoshan mine, South China. Sci. Total Environ. 407, 1551–1561. https://doi.org/10.1016/j.scitotenv.2008.10.061 (2009).CAS 
Article 
PubMed 

Google Scholar 
Hosseini, M., Nabavi, S. M. B., Nabavi, S. N. & Pour, N. A. Heavy metals (Cd Co, Cu, Ni, Pb, Fe, and Hg) content in four fish commonly consumed in Iran: Risk assessment for the consumers. Environ. Monit. Assess. 187, 237. https://doi.org/10.1007/s10661-015-4464-z (2015).CAS 
Article 
PubMed 

Google Scholar 
Prabhakaran, K., Nagarajan, R., MerlinFranco, F. & AnandKumar, A. Biomonitoring of Malaysian aquatic environments: A review of status and prospects. Ecohydrol. Hydrobiol. 17, 134–147. https://doi.org/10.1016/j.ecohyd.2017.03.001 (2017).Article 

Google Scholar 
Meche, A. et al. Determination of heavy metals by inductively coupled plasma-optical emission spectrometry in fish from the Piracicaba River in Southern Brazil. Microchem. J. 94, 171–174 (2010).CAS 

Google Scholar 
Zhang, Y. et al. Temporal and spatial changes of microbial community in an industrial effluent receiving area in Hangzhou Bay. J. Environ. Sci. 44, 57–68. https://doi.org/10.1016/j.jes.2015.11.023 (2016).CAS 
Article 

Google Scholar 
Huang, L. et al. Quantifying the spatiotemporal dynamics of industrial land uses through mining free access social datasets in the Mega Hangzhou Bay Region, China. Sustainability 10, 3463 (2018).
Google Scholar 
Pang, H.-J. et al. Contamination, distribution, and sources of heavy metals in the sediments of Andong tidal flat, Hangzhou bay, China. Continental Shelf Res. 110, 72–84. https://doi.org/10.1016/j.csr.2015.10.002 (2015).Article 

Google Scholar 
National Bureau of Statstics. Zhejiang Statistical Yearbook-2017 (China Statistics Press, 2017).
Google Scholar 
Chen, W., Zheng, Y., Chen, Y. & Mathews, C. An assessment of fishery yields from the East China Sea ecosystem. Mar. Fish. Rev. 59, 1–7 (1997).
Google Scholar 
Zhejiang Provincial Development and Reform Commission. Zhejiang Zhoushan Islands New Area Development Plan (In Chinese). (2021).Che, Y., He, Q. & Lin, W.-Q. The distributions of particulate heavy metals and its indication to the transfer of sediments in the Changjiang Estuary and Hangzhou Bay, China. Mar. Pollut. Bull. 46, 123–131 (2003).CAS 
PubMed 

Google Scholar 
Li, R. et al. Environmental health and ecological risk assessment of soil heavy metal pollution in the coastal cities of Estuarine Bay—a case study of Hangzhou Bay, China. Toxics 8, 75 (2020).CAS 
PubMed Central 

Google Scholar 
Bergami, E., Manno, C., Cappello, S., Vannuccini, M. L. & Corsi, I. Nanoplastics affect moulting and faecal pellet sinking in Antarctic krill (Euphausia superba) juveniles. Environ. Int. 143, 105999. https://doi.org/10.1016/j.envint.2020.105999 (2020).CAS 
Article 
PubMed 

Google Scholar 
Fang, H., Huang, L., Wang, J., He, G. & Reible, D. Environmental assessment of heavy metal transport and transformation in the Hangzhou Bay, China. J. Hazard. Mater. 302, 447–457 (2016).CAS 
PubMed 

Google Scholar 
Zhu, G. et al. Evaluation of ecosystem health and potential human health hazards in the Hangzhou Bay and Qiantang Estuary region through multiple assessment approaches. Environ. Pollut. 264, 114791. https://doi.org/10.1016/j.envpol.2020.114791 (2020).CAS 
Article 
PubMed 

Google Scholar 
Li, F. et al. Distribution and risk assessment of trace metals in sediments from Yangtze River estuary and Hangzhou Bay, China. Environ. Sci. Pollut. Res. 25, 855–866. https://doi.org/10.1007/s11356-017-0425-0 (2018).CAS 
Article 

Google Scholar 
Liu, L., Huang, X., Cao, W. & Yang, Y. Pollution load characteristics of the Hangzhou Bay and its surrounding areas. Ocean Dev. Manage 5, 108–112 (2012).
Google Scholar 
He, Z., Li, F., Dominech, S., Wen, X. & Yang, S. Heavy metals of surface sediments in the Changjiang (Yangtze River) Estuary: Distribution, speciation and environmental risks. J. Geochem. Explor. 198, 18–28. https://doi.org/10.1016/j.gexplo.2018.12.015 (2019).CAS 
Article 

Google Scholar 
Jin, X., Zhao, X., Meng, T. & Cui, Y. The Fishery Resources and the Environment of the Bohai Sea and Yellow Sea (Science Press, 2005).
Google Scholar 
Huang, Z. The Species and Distribution of Marine Organisms of China (Ocean Press, Beijing, 1994) (In Chinese).
Google Scholar 
Schram, F. R. Checklist of Marine Biota of China Seas. J. Crustac. Biol. 30, 339–339. https://doi.org/10.1651/09-3228.1 (2010).Article 

Google Scholar 
AQSIQ, P. in GB 17378.6–2007 (General Administration of Quality Supervision, Inspection and Quarantine of People’s Republic of China, 2007).Zhang, L. et al. Distribution and bioaccumulation of heavy metals in marine organisms in east and west Guangdong coastal regions, South China. Mar. Pollut. Bull. 101, 930–937. https://doi.org/10.1016/j.marpolbul.2015.10.041 (2015).CAS 
Article 
PubMed 

Google Scholar 
Zhong, W. et al. Health risk assessment of heavy metals in freshwater fish in the central and eastern North China. Ecotoxicol. Environ. Saf. 157, 343–349. https://doi.org/10.1016/j.ecoenv.2018.03.048 (2018).CAS 
Article 
PubMed 

Google Scholar 
Wang, Q. et al. Bioaccumulation and biomagnification of emerging bisphenol analogues in aquatic organisms from Taihu Lake, China. Sci. Total Environ. 598, 814–820. https://doi.org/10.1016/j.scitotenv.2017.04.167 (2017).CAS 
Article 
PubMed 

Google Scholar 
Arnot, J. A. & Gobas, F. A. A review of bioconcentration factor (BCF) and bioaccumulation factor (BAF) assessments for organic chemicals in aquatic organisms. Environ. Rev. 14, 257–297 (2006).CAS 

Google Scholar 
Duan, X., Zhao, X., Wang, B., Chen, Y. & Cao, S. Exposure Factors Handbook of Chinese Population (Adults) (China Environmental Science Press, 2013).
Google Scholar 
Chauhan, G. & Chauhan, U. Human health risk assessment of heavy metals via dietary intake of vegetables grown in wastewater irrigated area of Rewa, India. Int. J. Sci. Res. Publ. 4, 1–9 (2014).
Google Scholar 
USEPA. (Philadelphia PA; Washington, DC, 2007).Wang, X., Sato, T., Xing, B. & Tao, S. Health risks of heavy metals to the general public in Tianjin, China via consumption of vegetables and fish. Sci. Total Environ. 350, 28–37. https://doi.org/10.1016/j.scitotenv.2004.09.044 (2005).CAS 
Article 
PubMed 

Google Scholar 
USEPA. (2015).FAO/WHO. Wastewater Use in Agriculture. 988 (World Health Organization).Ahmed, A. S. S. et al. Bioaccumulation of heavy metals in some commercially important fishes from a tropical river estuary suggests higher potential health risk in children than adults. PLoS One 14, e0219336. https://doi.org/10.1371/journal.pone.0219336 (2019).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Saha, N., Mollah, M. Z. I., Alam, M. F. & Safiur Rahman, M. Seasonal investigation of heavy metals in marine fishes captured from the Bay of Bengal and the implications for human health risk assessment. Food Control 70, 110–118. https://doi.org/10.1016/j.foodcont.2016.05.040 (2016).CAS 
Article 

Google Scholar 
Yin, S., Feng, C., Li, Y., Yin, L. & Shen, Z. Heavy metal pollution in the surface water of the Yangtze Estuary: A 5-year follow-up study. Chemosphere 138, 718–725. https://doi.org/10.1016/j.chemosphere.2015.07.060 (2015).CAS 
Article 
PubMed 

Google Scholar 
USEPA. Risk-based concentration table. United States Environmental Protection Agency, Washington DC, Philadelphia (2000).Hu, B. et al. Assessment of heavy metal pollution and health risks in the soil-plant-human system in the Yangtze River Delta, China. Int. J. Environ. Res. Public Health 14, 1042 (2017).PubMed Central 

Google Scholar 
USEPA. in United States Environmental Protection Agency, Washington DC, Philadelphia (2010).Kwok, C. K. et al. Bioaccumulation of heavy metals in fish and Ardeid at Pearl River Estuary, China. Ecotoxicol. Environ. Saf. 106, 62–67. https://doi.org/10.1016/j.ecoenv.2014.04.016 (2014).CAS 
Article 
PubMed 

Google Scholar 
Yu, T., Zhang, Y., Hu, X. & Meng, W. Distribution and bioaccumulation of heavy metals in aquatic organisms of different trophic levels and potential health risk assessment from Taihu lake, China. Ecotoxicol. Environ. Saf. 81, 55–64. https://doi.org/10.1016/j.ecoenv.2012.04.014 (2012).CAS 
Article 

Google Scholar 
Qiu, Y.-W., Lin, D., Liu, J.-Q. & Zeng, E. Y. Bioaccumulation of trace metals in farmed fish from South China and potential risk assessment. Ecotoxicol. Environ. Saf. 74, 284–293. https://doi.org/10.1016/j.ecoenv.2010.10.008 (2011).CAS 
Article 
PubMed 

Google Scholar 
Arulkumar, A., Paramasivam, S. & Rajaram, R. Toxic heavy metals in commercially important food fishes collected from Palk Bay, Southeastern India. Mar. Pollut. Bull. 119, 454–459. https://doi.org/10.1016/j.marpolbul.2017.03.045 (2017).CAS 
Article 
PubMed 

Google Scholar 
Jonathan, M. P. et al. Metal concentrations in demersal fish species from Santa Maria Bay, Baja California Sur, Mexico (Pacific coast). Mar. Pollut. Bull. 99, 356–361. https://doi.org/10.1016/j.marpolbul.2015.07.032 (2015).CAS 
Article 
PubMed 

Google Scholar 
Liu, H., Yang, J. & Gan, J. Trace element accumulation in bivalve mussels Anodonta woodiana from Taihu Lake, China. Arch. Environ. Contam. Toxicol. 59, 593–601. https://doi.org/10.1007/s00244-010-9521-6 (2010).CAS 
Article 
PubMed 

Google Scholar 
Wang, W. X. et al. Copper and zinc contamination in oysters: Subcellular distribution and detoxification. Environ. Toxicol. Chem. 30, 1767–1774 (2011).CAS 
PubMed 

Google Scholar 
de FreitasRebelo, M., do Amaral, M. C. R. & Pfeiffer, W. C. High Zn and Cd accumulation in the oyster Crassostrea rhizophorae, and its relevance as a sentinel species. Mar. Pollut. Bull. 46, 1354–1358 (2003).
Google Scholar 
AQSIQ, P. in GB 18421–2001 (General administration of quality supervision, inspection and quarantine of People’s Republic of China, 2001).FAO/WHO. in Fifth Session [displayed 10 February 2014]. ftp://ftp.fao.org/codex/meetings/CCCF/cccf5/cf05_INF.pdf.Nauen, C. E. Compilation of legal limits for hazardous substances in fish and fishery products. FAO Fisheries Circular (FAO). no. 764. (1983).Rajeshkumar, S. & Li, X. Bioaccumulation of heavy metals in fish species from the Meiliang Bay, Taihu Lake, China. Toxicol. Rep. 5, 288–295. https://doi.org/10.1016/j.toxrep.2018.01.007 (2018).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Baki, M. A. et al. Concentration of heavy metals in seafood (fishes, shrimp, lobster and crabs) and human health assessment in Saint Martin Island, Bangladesh. Ecotoxicol. Environ. Saf. 159, 153–163. https://doi.org/10.1016/j.ecoenv.2018.04.035 (2018).CAS 
Article 
PubMed 

Google Scholar 
Vu, C. T., Lin, C., Yeh, G. & Villanueva, M. C. Bioaccumulation and potential sources of heavy metal contamination in fish species in Taiwan: Assessment and possible human health implications. Environ. Sci. Pollut. Res. 24, 19422–19434. https://doi.org/10.1007/s11356-017-9590-4 (2017).CAS 
Article 

Google Scholar 
Sharma, B., Singh, S. & Siddiqi, N. J. Biomedical implications of heavy metals induced imbalances in redox systems. BioMed Res. Int. 20, 14 (2014).
Google Scholar 
Feng, W., Wang, Z., Xu, H., Chen, L. & Zheng, F. Trace metal concentrations in commercial fish, crabs, and bivalves from three lagoons in the South China Sea and implications for human health. Environ. Sci. Pollut. Res. https://doi.org/10.1007/s11356-019-06712-8 (2020).Article 

Google Scholar 
Ruiz-Fernández, A. C. et al. A comparative study on metal contamination in Estero de Urias lagoon, Gulf of California, using oysters, mussels and artificial mussels: Implications on pollution monitoring and public health risk. Environ. Pollut. 243, 197–205 (2018).PubMed 

Google Scholar 
Bergstad, O. A. In Encyclopedia of Ocean Sciences (Second Edition) (ed. Steele, J. H.) 458–466 (Academic Press, 2009).
Google Scholar 
Mauchline, J. & Gordon, J. Foraging strategies of deep-sea fish. Mar. Ecol. Prog. Ser. 27, 227–238 (1986).
Google Scholar 
Li, J., He, M., Han, W. & Gu, Y. Analysis and assessment on heavy metal sources in the coastal soils developed from alluvial deposits using multivariate statistical methods. J. Hazard. Mater. 164, 976–981. https://doi.org/10.1016/j.jhazmat.2008.08.112 (2009).CAS 
Article 
PubMed 

Google Scholar 
Yu, P. Applications of hierarchical cluster analysis (CLA) and principal component analysis (PCA) in feed structure and feed molecular chemistry research, using synchrotron-based Fourier transform infrared (FTIR) microspectroscopy. J. Agric. Food Chem. 53, 7115–7127 (2005).CAS 
PubMed 

Google Scholar 
Kara, D. Evaluation of trace metal concentrations in some herbs and herbal teas by principal component analysis. Food Chem. 114, 347–354 (2009).CAS 

Google Scholar 
Chai, X. et al. Distribution, sources and assessment of heavy metals in surface sediments of the Hangzhou Bay and its adjacent areas. Acta Sci. Circum. 35, 3906–3916 (2015).CAS 

Google Scholar 
Mackay, D. & Fraser, A. Bioaccumulation of persistent organic chemicals: Mechanisms and models. Environ. Pollut. 110, 375–391. https://doi.org/10.1016/S0269-7491(00)00162-7 (2000).CAS 
Article 
PubMed 

Google Scholar 
ATSDR, T. ATSDR (Agency for toxic substances and disease registry). Prepared by Clement International Corp., under contract 205, 88–0608 (2000).Traina, A. et al. Heavy metals concentrations in some commercially key species from Sicilian coasts (Mediterranean Sea): Potential human health risk estimation. Ecotoxicol. Environ. Saf. 168, 466–478. https://doi.org/10.1016/j.ecoenv.2018.10.056 (2019).CAS 
Article 
PubMed 

Google Scholar 
Ozmen, M., Ayas, Z., Güngördü, A., Ekmekci, G. F. & Yerli, S. Ecotoxicological assessment of water pollution in Sariyar Dam Lake, Turkey. Ecotoxicol. Environ. Saf. 70, 163–173. https://doi.org/10.1016/j.ecoenv.2007.05.011 (2008).CAS 
Article 
PubMed 

Google Scholar 
Jeffrey, B. & Alison, G. Guidance for assessing chemical contaminant data for use in fish advisories. v. 1. Fish sampling and analysis-v. 4. Risk communication. (1993).Regulations, U. S. E. P. A. O. o. W. Assessing Human Health Risks from Chemically Contaminated Fish and Shellfish: A Guidance Manual. (US Environmental Protection Agency, 1989).Liu, Q., Liao, Y. & Shou, L. Concentration and potential health risk of heavy metals in seafoods collected from Sanmen Bay and its adjacent areas, China. Mar. Pollut. Bull 131, 356–364. https://doi.org/10.1016/j.marpolbul.2018.04.041 (2018).CAS 
Article 
PubMed 

Google Scholar 
Abtahi, M. et al. Heavy metals (As, Cr, Pb, Cd and Ni) concentrations in rice (Oryza sativa) from Iran and associated risk assessment: A systematic review. Toxin Rev. 36, 331–341 (2017).CAS 

Google Scholar 
WHO. WHO Technical Report Series. Evaluation of Certain Food Additives and Contaminants. Fifty-Third Report of the Joint FAO/WHO Expert Committee on Food Additives (JECFA). http://www.Who.Int/foodsafety/publications/jecfa-reports/en/ (2000).USEPA. USEPA Regional Screening Level (RSL) summary table: November 2011. (2011).Farkas, A., Salánki, J. & Specziár, A. Age-and size-specific patterns of heavy metals in the organs of freshwater fish Abramis brama L. populating a low-contaminated site. Water Res. 37, 959–964 (2003).CAS 
PubMed 

Google Scholar 
Canpolat, Ö. & Çalta, M. Heavy metals in some tissues and organs of Capoeta capoeta umbla(Heckel, 1843) fish species in relation to body size, age, sex and seasons. Fresenius Environ. Bull. 12, 961–966 (2003).CAS 

Google Scholar 
Hosseini, M., Nabavi, S. M. B., Nabavi, S. N. & Pour, N. A. Heavy metals (Cd Co, Cu, Ni, Pb, Fe, and Hg) content in four fish commonly consumed in Iran: Risk assessment for the consumers. Environ. Monit. Assess. 187, 1–7 (2015).CAS 

Google Scholar 
Jiang, X. et al. Assessment of heavy metal accumulation in freshwater fish of Dongting Lake, China: Effects of feeding habits, habitat preferences and body size. J. Environ. Sci. 112, 355–365 (2022).
Google Scholar 
Yi, Y., Tang, C., Yi, T., Yang, Z. & Zhang, S. Health risk assessment of heavy metals in fish and accumulation patterns in food web in the upper Yangtze River, China. Ecotoxicol. Environ. Saf. 145, 295–302 (2017).CAS 
PubMed 

Google Scholar 
USEPA. Assessing Human Health Risks from Chemically Contaminated Fish and Shellfish: A Guidance Manual. (US Environmental Protection Agency, 1989).Means, B. Risk-assessment guidance for superfund. Volume 1. Human health evaluation manual. Part A. Interim report (Final). (Environmental Protection Agency, Washington, DC (USA). Office of Solid Waste …, 1989).Raknuzzaman, M. et al. Trace metal contamination in commercial fish and crustaceans collected from coastal area of Bangladesh and health risk assessment. Environ. Sci. Pollut. Res. 23, 17298–17310. https://doi.org/10.1007/s11356-016-6918-4 (2016).CAS 
Article 

Google Scholar 
Kalantzi, I. et al. Metals in tissues of seabass and seabream reared in sites with oxic and anoxic substrata and risk assessment for consumers. Food Chem. 194, 659–670. https://doi.org/10.1016/j.foodchem.2015.08.072 (2016).CAS 
Article 
PubMed 

Google Scholar 
Sarkar, S., Mukherjee, S., Chattopadhyay, A. & Bhattacharya, S. Differential modulation of cellular antioxidant status in zebrafish liver and kidney exposed to low dose arsenic trioxide. Ecotoxicol. Environ. Saf. 135, 173–182. https://doi.org/10.1016/j.ecoenv.2016.09.025 (2017).CAS 
Article 
PubMed 

Google Scholar 
Mandal, B. K. & Suzuki, K. T. Arsenic round the world: A review. Talanta 58, 201–235. https://doi.org/10.1016/S0039-9140(02)00268-0 (2002).CAS 
Article 
PubMed 

Google Scholar 
Kibria, G., Hossain, M. M., Mallick, D., Lau, T. C. & Wu, R. Trace/heavy metal pollution monitoring in estuary and coastal area of Bay of Bengal, Bangladesh and implicated impacts. Mar. Pollut. Bull. 105, 393–402. https://doi.org/10.1016/j.marpolbul.2016.02.021 (2016).CAS 
Article 
PubMed 

Google Scholar 
Fang, Y. et al. Concentrations and health risks of lead, cadmium, arsenic, and mercury in rice and edible mushrooms in China. Food Chem. 147, 147–151. https://doi.org/10.1016/j.foodchem.2013.09.116 (2014).CAS 
Article 
PubMed 

Google Scholar 
Vannoort, R. & Thomson, B. New Zealand Total Diet Study—Agricultural Compound Residues (Selected Contaminant and Nutrient Elements. Ministry for Primary Industries, 2009).
Google Scholar 
Praveena, S. M., Pradhan, B. & Ismail, S. N. S. Spatial assessment of heavy metals in surface soil from Klang District (Malaysia): An example from a tropical environment. Hum. Ecol. Risk Assess. Int. J. 21, 1980–2003 (2015).CAS 

Google Scholar  More

Study sitesWe sampled foliar fungal endophytes and root fungi (root endophytes and AM fungi) in the Colorado Rockies at the Rocky Mountain Biological Laboratory, Gunnison Co., Colorado, USA (38°57’N, 106°59’W). This region has predictable decreases in air temperature (c. 0.8 °C per 100 m; [40]) and declines in soil nutrients with altitude [41], but increases in precipitation, mainly as snow [42]. The entire region is warming at rates of 0.5–1.0 °C per decade [43].To capture environmental, spatial, and grass-host specific variation in fungal guilds, we sampled 66 sites encompassing 9–13 elevations from each of six altitudinal gradients in July 2014 (Supplementary Table S1, Supplementary Fig. S1). Elevational gradients represented separate mountains in the Gunnison Basin and were located within 20 km of each other. We created a regional climate model to interpolate average number of growing degree days (GDD, base 0 °C), mean annual temperature (MAT), maximum temperature (Tmax), minimum temperature (Tmin), mean annual precipitation (MAP), and mean snow depth (MSD) for each site based on data from 29 local meteorological stations [44]. At each site, soil edaphic parameters were measured on dried soil at the UC Davis soils lab (see [24] for more details) and soil nutrients at Western Ag (Saskatoon, Canada). Soil pH was measured in a 1:1 solution with diH2O, and soil moisture was measured gravimetrically. Physical characteristics of each site (e.g., aspect, soil depth, elevation) were measured as described in Lynn et al. [44]. Environmental variation across sites was large. For example, MAT varied from 7.1 to 13.3 °C, MAP from 563 to 1171 mm, and Total N from 2 to 316 ug/g dry soil (Table S1).Host plant speciesWe focused on grasses because grasslands cover ~20% of Earth’s land surface [45] and dominate subalpine meadows of the Rocky Mountains. In addition, individual grass species spanned the entire elevational range of our study system [46], whereas tree, shrub, and forb species did not. At each location, we sampled nine adult individuals from up to 13 grass species representing five genera (Poaceae, subfamily Pooideae; Supplementary Table S1). Many sites had fewer than 13 grass species present, but all sites, except for two, had at least two grass species. Samples were composited by tissue type (leaves v. roots) and grass species within each site.Fungal compositionCollected root and leaf samples were surface sterilized (1 min in 95% ethanol, 2 min in 1% sodium hypochlorite solution, and 2 min in 70% ethanol) over ice to focus on the endophytic fungal community [34]. Following surface sterilization, samples were rinsed in purified water (Milli-Q Integral Water Purification System, EMD Millipore Corporation, Billerica, MA), stored in RNAlater, and refrigerated. All samples were then frozen in liquid nitrogen and ground using a mortar and pestle. Total DNA was extracted from ~50 mg of ground sample using QIAGEN DNeasy plant extraction kits (QIAGEN Inc., Valencia, CA).Fungal composition was characterized using barcoded primers targeting the ITS2 region for leaf and root endophytes [47], and FLR3-FLR4 primers targeting ~300 bp in the 28S region for AMF [48]. Each PCR contained 5 μL of ~1–10 ng/μL DNA template, 21.5 μL of Platinum PCR SuperMix (Thermo Fisher Scientific Inc., Waltham, MA), 1.25 μL of each primer (10 μM), 1.25 μL of 20 mg/mL BSA, and 0.44 μL of 25 mM MgCl2. For the ITS2 primers, the reactions included an initial denaturing step at 96 °C for 2 min, followed by 24 cycles of 94 °C for 30 sec, 51 °C for 40 s, and 72 °C for 2 min, with a final extension at 72 °C for 10 min. For the 28S primers, reactions started with an initial denaturing step at 93 °C for 5 min, followed by 33 cycles of 93 °C for 1 min, 55 °C for 1 min, and 72 °C for 1 min, with a final extension at 72 °C for 10 min.Three PCR replicates from each sample were pooled and then cleaned and concentrated using a ZR-96 DNA Clean & Concentrator-5 (Zymo Research Corporation, Irvine, CA). PCR was then carried out on all samples to add dual indexes and Illumina sequencing adaptors; each reaction began with an initial denaturing step at 98 °C for 30 s, followed by 7 cycles of 98 °C for 30 s, 62 °C for 30 s, and 72 °C for 30 s, with a final extension at 72 °C for 5 min. Sequencing was performed by the Genomic Sequencing and Analysis Facility at The University of Texas at Austin using paired-end 250 base Illumina MiSeq v.3 chemistry (Illumina, Inc., San Diego, CA). We aimed to obtain a minimum of 30,000 reads/sample for the ITS2 region and 20,000 reads/sample for the 28S region. All sequences are deposited in the NCBI SRA database under accession number (PRJNA639093).BioinformaticsWe processed reads to generate OTUs using commands from USEARCH (v9.2.64). Reads from previous studies [24] and this study were clustered together to improve OTU delineations for a total of 36,754,931 reads. We merged paired-end reads using the fastq_mergepairs from USEARCH with “fastq_maxdiffs” set to 20 and “fastq_maxdiffpct” set to 10 to ensure proper merging at a low error rate. The merged reads and the forward unmerged reads were trimmed at the primer sites using cutadapt with “e” set to 0.2, “m” set to 200, and untrimmed reads were discarded. Merged reads were filtered using fastq_filter from USEARCH with “fastq_maxee” set to 1.0. The forward reads were first trimmed to 230 using fastx_truncate from USEARCH with “trunclen” set to 230 and then filtered by fastq_filter from USEARCH with “fastq_maxee” set to 1.0. We then concatenated the merged and forward reads into one file and de-replicated using fastx_uniques from USEARCH with “minuniquesize” set to 2. After these steps, 11,357,274 sequences remained. We clustered these sequences to form OTUs at 97% similarity [49] using cluster_otus command from UPARSE. The reads (all reads before filtering step) of each sample were mapped to OTUs with usearch_global from USEARCH with “id” set to 0.97. We determined taxonomy for the representative OTUs using sintax from USEARCH with the database set to UNITE all eukaryotes (v. 8.2) “strand” set to both and “sintax_cutoff” set to 0.8 [50]. Representative OTUs were also blasted against Genbank with “perc_identity” set to 80 and “max_target_seqs” set to 50. All OTUs identified as “fungi” were retained, and OTUs labeled as “unknown” or “unidentified” were manually inspected based on blast results to determine retention. Our filtering criteria left between 5 and 418 OTUs per sample (Supplementary Table S2).Due to low fungal abundance in leaves [34], many leaf samples were dominated by plant sequences (average ~78% plant reads). Therefore, fungal sequence numbers in leaf samples were low, despite adequate sequencing depth to capture trends in fungal endophyte communities across sites based on prior analyses [24, 34, 35]. We included only samples that contained at least 50 fungal sequences after data processing (Leaves N = 192, Roots N = 191, AMF N = 251), and most samples had much greater sequencing depth, especially for roots (Supplementary Table S2). Nevertheless, there were no correlations between sequence read depth and richness, alpha diversity, or evenness of our samples (P  > 0.05 in all cases), and plant species did not differ in the average sequencing depth for samples (P  > 0.05). Data for each fungal OTU were transformed to the proportion of total sequence abundance to minimize any differences in sampling effort [51].Diversity and compositionWe calculated the alpha diversity metrics of richness, Shannon’s Diversity, Inverse Simpson’s Diversity, and Pielou’s Evenness. For each fungal guild, differences among plant species and elevation in alpha diversity were first determined using a general linear mixed effects model with plant species (categorical) and elevation (continuous) as fixed effects and site nested within elevation gradient (e.g., mountain identity, Supplementary Table S1, Supplementary Fig. S1) as random effects to account for the lack of statistical independence among plant species sampled at the same site and among sites located within the same mountain elevation gradient (Supplementary Fig. S1). Models were constructed using the lmer function in R package lme4 [52, 53]. To address, do fungal community patterns along environmental gradients differ among guilds: leaf endophytes, root endophytes, or arbuscular mycorrhizal fungi?, we then compared alpha diversity metrics among fungal guilds using a general linear mixed effects model with fungal guild, plant species, and elevation as fixed effects and site nested within elevation gradient as random effects. In all models, we evaluated parameter fit with analysis of deviance using Wald chi-square tests and corrected for multiple comparisons using a false discovery alpha of 0.05. Differences among grass species were determined using Tukey post-hoc tests.Because elevation is a good proxy for variation in both climate and soil parameters (Supplementary Table S1), in all community analyses, we first ran models with grass species and elevation to parse biotic versus abiotic influences on fungal OTUs, then secondly ran full variance partitioning models with all environmental covariates (Supplementary Table S1, climate, physical, soil) in addition to grass species identity and space (gradient location, Supplementary Fig. S1). Because leaf and root endophytes were sequenced using different primers than AM fungi, we could not compare composition among the three guilds directly. Instead, we compared the relative influence of biotic and abiotic drivers on fungal composition within each guild to compare patterns among guilds. To do so, we first used distance-based redundancy analysis (dbRDA) to analyze the effects of plant host species and elevation on fungal composition for general fungal communities in leaves and roots and separately for AM fungal communities in roots. All models were run on quantitative Jaccard indices of fungal composition for each guild and included site nested within elevation gradient (e.g., mountain side, Supplementary Fig. S1) as random effects. Second, to evaluate which environmental variables most strongly influenced fungal composition, we further partitioned variance in fungal composition due to grass species, climate variables (MAP, MAT, MSD, Tmax, Tmin, and GDD), soil variables (total nitrogen, total phosphorus, nitrate, ammonium, calcium, magnesium, potassium, iron, manganese, sulfur, aluminum, soil pH, soil gravimetric moisture content), physical variables (aspect degree, aspect category (e.g., cardinal direction), slope, soil depth, and elevation) and spatial variables (latitude and longitude) using the varpart function in Vegan v. 2–5.3 [54]. Plots of fungal composition by plant host were also generated using dbRDA separately for each fungal guild. Spatial variables were de-trended and tested for spatial autocorrelation using the ade4 package v. 1.7–16 [55]. When we detected significant spatial autocorrelation eigenvectors, we included these in the spatial variable matrix. To characterize how many fungal taxa occurred in multiple plant taxa and elevations, we used the VennDiagram package v. 1.6.20 [56].Turnover and rewiringTo evaluate whether fungal composition was driven by grasses associating with different fungal taxa or differing relative abundances of the same fungal taxa, we first performed a beta partitioning analysis using betapart v. 1.5.3 [57]. Each fungal guild was analyzed separately. Next, to examine turnover in the abundances of fungal functional groups (pathogens, saprotrophs, mutualists), we defined groups using the FungalTrait database, which merges previous databases into one cohesive framework of 17 functional trait types (referred to here as functional groups; [58]). We recognize that fungal functions are highly environmentally dependent and therefore these functional groups may represent potential function more than actual function. Functional group identity was ascribed to 60% of leaf endophyte and 62% of root endophyte fungal taxa. Then, cumulative abundance of proportionally transformed sequence reads in each functional group was analyzed using a general linear mixed effects model with grass species and elevation as fixed effects and site nested within elevation gradient as random effects, as above. Finally, we defined indicator species within the OTUs that comprised at least 1% of the total abundance of each fungal guild by grass host, gradient, and elevation classes (rounded to the nearest 100 m) using the indicspecies package v. 1.7.9 [59]. Functional group assignments using the FungalTrait database from above were assigned to each indicator taxon [58]. A large percentage of significant indicator taxa out of the total number of OTUs would confirm that turnover in the species identity of fungal associations is stronger than turnover in the relative abundances of the same fungal taxa.Network propertiesTo address does grass-fungal network structure track elevation?, we analyzed four properties that encompass different facets of ecological networks at the site level. First, we calculated network nestedness, or the propensity for specialists to interact with the same plant species as generalists, using the weighted NODF (Nestedness metric based on Overlap and Decreasing Fill; [60]). Second, we calculated complexity as linkage density or the average number of interactions per plant species [61]. Third, to characterize specialization, we used the H2’ Index [62]. Finally, network evenness was calculated as Alatalo’s interaction evenness [63]. In all cases, these network metrics were weighted indices to increase accuracy [64], and calculations were performed in the Bipartite package v. 2.15 [65]. To address, how much do fungal guilds differ in altitudinal variation in network structure?, we compared network-level statistics among fungal guilds using a general linear mixed effects model with fungal guild as a fixed effect, number of grass hosts as a fixed effect, and gradient as a random effect (function lmer in lme4 [52],). We compared relationships with elevation separately for each fungal guild, using general linear mixed effects models with elevation as a fixed, continuous effect, number of grass hosts within the network as a fixed, continuous effect, and gradient identity as a random effect (Supplementary Table S1, Supplementary Fig. S2). We evaluated parameter fit with analysis of deviance using Wald chi-square tests using the car package 3.0–10 in R [66].All data met model assumptions of normality of residuals and homogeneity of variance. All analyses were performed in R v. 3.5.0 [53]. More

The Bayesian VMD Method we developed can classify pulsed signals with similar frequency content in poor SNR files from underwater acoustic recordings. The Method consists of two parts. The first part scans the incoming audio data as segments that potentially contain signals of interest by detecting energy peaks. It then uses the start and end of the energy peaks to isolate those areas of interest from non-signal areas of the audio file. The second part classifies the detected signals into separate categories based on their frequency content. The algorithms of our Detector and Classifier steps are self-developed, but some key components in them were inspired by previous work39,40,41.DetectorThe proposed detector uses full audio files that are 4.5 s long at a sampling rate of 100 kHz. It then finds audio file segments where potential signals of interest exist.For a given audio file, denoted by ({hat{x}}(n)), where (n=1, dots , N), and N is the total number of samples, the Laplacian Differential Operator (LDO) is applied to ({hat{x}}(n)) resulting in an enhanced version of the audio file denoted by y(n), as follows:$$begin{aligned} y(n) = frac{1}{4}frac{partial ^2 {hat{x}}}{partial n^2} end{aligned}$$
(1)
The LDO enhances the transient signals (edge detection) and filters out the low frequencies ((< 10) kHz) which are not needed for Gg and Lo pulsed signal classification. The y(n) is then transformed into a time-frequency representation using Short-time Fourier transform (STFT). The STFT was implemented on 1024 samples with 90% overlap and a 1024-point Hanning window. The magnitude of the STFT matrix s(n, f) is given as ({hat{S}}(n,f)).$$begin{aligned} {hat{S}}(n,f) = begin{bmatrix} |s_{11}| &{} dots &{} |s_{1N}|\ vdots &{} ddots \ |s_{M1}| &{} &{} |s_{MN}| end{bmatrix} end{aligned}$$ (2) Where (N) is the length of the input segment and (M) is the number of frequency bins. The dimensionality of matrix ({hat{S}}(n,f)) is reduced from 2-D to 1-D as follows:$$begin{aligned} S_{d}(n) = sum _{f=1}^{M} {hat{S}}(n,f) end{aligned}$$ (3) The resulting temporal sequence is an accumulated sum of all frequency bins from (begin{aligned} {hat{S}}(n,f) end{aligned}), so scaling is applied, as follows:$$begin{aligned} S_{d}(n) = frac{S_{d}(n)}{max{S_{d}(n)}} end{aligned}$$ (4) After finding (S_{d}(n)) from Eq. (4), the mean of (S_{d}(n)) is subtracted. Then, to determine the boundaries of the acoustic signal, an adaptive threshold is applied. The first step in developing the threshold is to vectorize the matrix ({hat{S}}(n,f)) in column order into a vector called (S_{r}(n)):$$begin{aligned} S_{r}(n) = overrightarrow{{hat{S}}(n,f)} end{aligned}$$ (5) Then, (S_r (n)) is scaled similar to (S_{d}(n)) and is sorted into ascending order, denoted by ({hat{S}}_r(n)). The changing point where the root-mean-square level of the sorted curve ({hat{S}}_r(n)) changes the most is obtained by minimizing Eq. (6)39,40,42$$begin{aligned} J(k) = sum _{i=1}^{k-1} Delta ({hat{S}}_{r,i}; chi ([{hat{S}}_{r,1} dots {hat{S}}_{r,k-1}])) + sum _{i=k}^{N} Delta ({hat{S}}_{r,i}; chi ([{hat{S}}_{r,k} dots {hat{S}}_{r,N}])) end{aligned}$$ (6) where (k) and N are the index of the changing point and the length of the sorted curve ({hat{S}}_r (n)), respectively, and$$begin{aligned} sum _{i=u}^{v} Delta ({hat{S}}_{r,i}; chi ([{hat{S}}_{r,u} dots {hat{S}}_{r,v}])) = (u-v+1)log left( frac{1}{u-v+1}sum _{n=u}^{v}{hat{S}}_{r,n},^{2}right) end{aligned}$$ (7) The threshold, (lambda), is the value of ({hat{S}}_r (k)) which equals the noise floor estimation, and can be represented as follows:$$begin{aligned} begin{aligned} {mathcal {H}}_{0}: S_d(n) < lambda \ {mathcal {H}}_{1}: S_d(n) ge lambda end{aligned} end{aligned}$$ (8) where ({mathcal {H}}_0) and ({mathcal {H}}_1) are the hypothesis that the activity was below or above the threshold, respectively. The calculated threshold can vary for each file, thus making it adaptable if ambient noise conditions change between files. The threshold (lambda) is then projected onto the temporal sequence (S_{d}(n)) to extract the boundaries of the regions of the acoustic signal that comprised the detected energy peak. The start and end points of each acoustic signal are determined as the first and last points that are greater than (lambda) in amplitude.The boundaries of the detected segments are scaled by the sampling rate to obtain start and end times which will be used to extract the audio file segments from the original data file in the classification step. Figure 4 illustrates the layout of the the proposed detector.Figure 4Block diagram of the proposed detector.Full size imageClassifierOnce segments with energy peaks were identified, they were scanned by the team’s bioacoustics expert, and any segments confirmed to contain only Gg or Lo signals were sifted out for use in testing the accuracy of the Bayesian VMD Method classifier.In this paper, the metric weight was defined for classification purposes. The weight for a parameter (varvec{theta _i}) given its measurement (varvec{y_i}) is defined as$$begin{aligned} w(varvec{theta _i} mid varvec{y_i}) = P_{varvec{Theta mid Y}}(varvec{theta _i} mid varvec{y_i}) * varvec{p_i} end{aligned}$$ (9) where (varvec{theta _i}) is the probability density function (PDF) of (varvec{y_i}), (varvec{y_i}) is one measurement in the measurement vector (varvec{y}), (P_{varvec{Theta mid Y}}(varvec{theta _i} mid varvec{y_i})) is the posterior probability of the parameter (varvec{theta _i}) given the measurement (varvec{y_i}), and (varvec{p_i}) is the scaled prominence value of (varvec{y_i}).When a detected audio file segment is fed into the Bayesian VMD classifier, the classification process starts with a feature extraction step. During this step, peak and notch frequencies and their prominence values were obtained from the VMD-Hilbert spectrum of the segment. The prominence measures how much a peak stands out due to its intrinsic height or how much a notch stands out due to its depth and its location relative to surrounding peaks or notches. In general, peaks that are taller and more isolated have a higher “prominence” (p) than peaks that are shorter or surrounded by other peaks.In the feature extraction step, VMD decomposed the input audio segment into a set of IMFs. The HHT was then applied to all IMFs to create a Hilbert spectrum with a frequency resolution of 50 Hz. The Hilbert spectrum is a matrix, (H(n,f)) that contains the instantaneous energies, (h(n,f)).$$begin{aligned} H(n,f) = begin{bmatrix} h_{11} &{}dots &{} h_{1R} \ vdots &{} ddots \ h_{Q1} &{} &{} h_{QR} end{bmatrix} end{aligned}$$ (10) where r is the length of the input segment and q is the number of frequency bins in (H).The matrix (H (n,f)) is then converted from a 2-D array to a 1-D spectral representation by summing all instantaneous energy values in each frequency bin, as follows:$$begin{aligned} H(f) = sum _{n=1}^{R} H(n,f) end{aligned}$$ (11) The energy summation sequence was converted to a base-10 logarithmic scale and then smoothed by passing through a 17-point median filter and an 11-point moving average filter for the purpose of easily extracting features. All peaks and notches in the sequence whose prominence values exceeded the threshold of 0.5 were located, and their frequency values and prominence values were then stored as extracted features from the input signal (see Fig. 5).Figure 5Example of locating peak and notch frequencies and how prominent they are compared to other peaks and notches. The wave form in (a) is the smoothed energy summation sequence from the Hilbert spectrum of the Lo signal in Fig. 1. Subplot (b) is a flipped version of the energy summation sequence for the convenience of extracting notch frequencies and their prominence values. The length of the red line represents the prominence value of a peak or notch.Full size imageFor testing the effectiveness of the VMD feature extractor, a second set of features were extracted from the FFT-based power spectrum using the same input signals with the Welch’s algorithm. The FFT-based spectrum was calculated on 2048 samples with 50% overlap and a 2048-point Hanning window with 48.82 Hz frequency resolution. The power spectral density sequence was then converted to dB and went through a 21-point median filter and a 15-point moving average filter. Feature extraction followed the same strategies as in VMD feature extractor except using a prominence threshold of 2 dB.Next, the measured features, frequencies (Hz) of the peaks and notches (henceforth referred to as “measured peaks and notches”), were matched with the probability distribution functions (PDFs) of peaks and notches (henceforth referred to as “parameter peaks and notches”) from Soldevilla et al. (2008). The matching between measured and parameter peaks and notches was done in preparation of weight calculations, and it was implemented for both Gg and Lo. There are four Gaussian PDFs for parameter peaks and three for parameter notches for each species in Soldevilla et al. (2008) (Table 2). A 95% confidence interval of a Gaussian PDF was used here as a frequency range defined as 1.96 standard deviations to the left and right of its mean value. When measured peaks and notches were matched to parameter peaks and notches, only the peak or notch that fell within a 95% confidence interval were kept. Any peaks or notches outside the 95% confidence intervals were discarded.Because there are overlaps between the 95% confidence intervals of 22.4 kHz and 25.5 kHz parameter peaks of Gg and between 33.7 kHz and 37.3 kHz parameter peaks of Lo (see Table 2), it is likely that some measured peaks will fall in the overlapping areas. In this paper, the maximum a posterior (MAP) estimation41 was used to determine which PDF results in the measured peak in an overlapping area. For a measured peak that falls into an overlapping area, two parameter peaks’ PDFs are plugged in the MAP estimation equation sequentially, and then the measured peak will be matched with the PDF that maximizes the posterior probability of it given the measured peak.After the preliminary match, if more than one measured peak or notch remains in any one PDF confidence interval, the measured peak and notch with the highest prominence value is selected as the real measured peak or notch of this PDF, and the redundant ones are discarded. Finally, all remaining peak prominence values and notch prominence values were scaled to be between 0 and 1, respectively.Once peak and notch matching and selection was finished, Bayesian weights were calculated to select the most likely species. From Bayes’s rule, the posterior probability of a parameter given its measurement is proportional to the product of the likelihood function of the measurement given the parameter and the prior probability of the parameter41, as shown in Eq. (12).$$begin{aligned} P_{varvec{Theta mid Y}}(varvec{theta _i} mid varvec{y_i}) propto f_{varvec{Y mid Theta }}(varvec{y_i} mid varvec{theta _i}) P_{varvec{Theta }}(varvec{theta _i}) end{aligned}$$ (12) therefore, substitution of the posterior probability in Eq. (9) yields$$begin{aligned} w(varvec{theta _i} mid varvec{y_i}) = f_{varvec{Y mid Theta }}(varvec{y_i} mid varvec{theta _i}) *P_{varvec{Theta }}(varvec{theta _i}) * varvec{p_i} end{aligned}$$ (13) Figure 6Example of feature matching. The top plots show a set of measured peaks and notches matched with both Gg’s PDFs (a) and Lo’s PDFs (b) parameter peaks and notches like in Fig. 5 during the feature matching and selection step. Middle plots show how closely to the parameter PDFs that the measured peaks match either Gg (c) or Lo (d) and their weight calculations. The width of each PDF represents its 95% confidence interval, and the ordinate represents the weight value. Subplots (e) and (f) show the same weight calculations for notches. The final weight value is the summation of all weight values of peaks and notches matched with Gg or Lo.Full size imageWith all PDFs and a priori probabilities from Soldevilla et al. (2008), the weight value in terms of Gg and Lo given a set of measurements, (varvec{y}), was obtained by Eqs. (13) and (14)$$begin{aligned} w(Gg mid varvec{y}) = sum _{forall i} w(varvec{theta _i} mid varvec{y_i}) qquad w(Lo mid varvec{y}) = sum _{forall j} w(varvec{theta _j} mid varvec{y_j}) end{aligned}$$ (14) where (varvec{y_i}) and (varvec{y_j}) are the remaining measured peaks and notches that were matched with Gg’s PDFs and Lo’s PDFs after the matching and matching step. The feature matching and selection results and the weight calculation process are shown in Fig. 6.The last step was a comparison between weight values in terms of Gg and Lo. If (w(Lo mid varvec{y}) > w(Gg mid varvec{y})), the signal was labeled an Lo signal; otherwise, it was labeled a Gg signal. The classifier is illustrated in Fig. 7. The weight values are significant to three digits because weights are normally smaller than 1.000 and three significant digits was sufficient for comparing all calculated weight values for these audio files. In the case that the weight comparison is equal to three significant digits (even though this never happened in these 174 signals), the Bayesian VMD algorithm will automatically classify the input as a Gg signal given that the highest precision (85.91%) by the Bayesian VMD Method was achieved on Gg.Figure 7Block diagram of the Bayesian VMD Method classifier.Full size image More