Interacting effects of habitat structure and seeding with oysters on the intertidal biodiversity of seawalls

they contained habitat structure, seeded oysters or both suggests there was some redundancy in the effects of the treatments, or, alternatively, because oyster cover did not differ between the three treatments, that oyster cover was the driver of species density. Oysters may influence species density by: enhancing the surface area for attachment and grazing; by providing micro- habitats between their shells that are protected from extreme low-tide air temperatures, desic- cation stress and/or predators; through their biological functions of filter feeding, which can reduce the abundance of competitively dominant spores, or create currents that facilitate set- tlement; and through biodeposition which provides organic material [53–55]. The species den- sity of mobile invertebrates and cryptobenthic fishes, may have responded more strongly to the initial seeding treatment than sessile taxa as they can move onto tiles within hours of their deployment to graze and utilise the microhabitat. In contrast, the colonisation of sessile taxa is dependent on larval settlement and may take longer to establish. Yet despite the rapid response of mobile species density to seeding with oysters, their abundances first displayed effects of habitat structure at 6 months and seeding at 12 months, Differences in the species density of sessile taxa among flat tiles and those with habitat structure may reflect the settlement choices of the organisms for cooler, shaded, moist and potentially less wave exposed surfaces [56],

greater post-settlement mortality on unprotected surfaces, or reduced competition for space where more hard substrate is available for settlement. Our experimental manipulations of seeded oysters were predicted to influence the species density and MaxN of pelagic fishes by enhancing prey availability and altering their foraging behaviour. However, contrary to our expectations, there were no temporally persistent effects of seeding on the species density or abundances of pelagic fishes. Previous research suggests that the relatively short-lived effect of seeding on the MaxN of pelagic fish, which disappeared after 6 months, was related to changing patterns of foraging on the tiles [7]. The most abun- dant pelagic fishes at 1 and 6 months were the Yellowfin bream (Acanthopagrus australis) and the Fanbelly leather jacket (Monacanthus chinensis). Both of these species are common around artificial habitats [57, 58] and are known predators of the Sydney rock oyster, S. glomerata [59–61], including those growing on the unprotected surfaces of seawalls [7, 62]. The absence of an effect of seeding on pelagic fishes after 6 months may reflect the loss of the oyster prey resource due to growth of oysters out of size classes most vulnerable to predation [7], and/or PLOS ONE Effects of microhabitats and seeding with oysters on intertidal biodiversity PLOS ONE | https://doi.org/10.1371/journal.pone.0230807 July 16, 2020 15 / 21

the complete removal of unprotected individuals from exposed surfaces, leaving only those in hard-to-get protected microhabitats. On seawalls, the distribution and abundances of organisms are often limited by abiotic and biotic stressors [15, 63–67]. There are increasing efforts to eco-engineer missing microhabitats on seawalls to enhance the biodiversity and ecosystems functioning of their communities (e.g. clearance rates, carbon sequestration or bioprotection), [23, 38, 42, 67–69]. Small-scale experi- ments such as that conducted here can be useful in identifying appropriate eco-engineering approaches before they are scaled up. Our study clearly demonstrates that crevices can enhance shading and moisture retention and reduced the high temperature extremes at low tides compared with ridges and flat tiles. They may also reduce fish predation on the colonising taxa, although this was not directly quantified in this study. Overall the presence of crevices increased the range of environmental conditions present on complex tiles which in turn could lead to greater co-existence of sessile and mobile taxa through portioning of resources [6]. In demonstrating that crevices support increased species density of sessile taxa and species density and abundances of mobile taxa as compared to ridges, our results support previous suggestions that the addition of crevices to homogenous, flat surfaces such as seawalls might be a successful ecoengineering approach [19]. Nevertheless, to fully assess the benefits, we would also need to compare the associated community of the

creviced tiles to those of featureless seawalls. Our research suggests that in addition to cervices, seeding with oysters influences the biodi- versity of the sessile taxa, mobile taxa and cryptobenthic fishes supported by artificial substrata [70, 71]. We demonstrated that the species density of sessile taxa benefited from adding either habitat structure or seeding with oysters, while the species density and/or abundances of mobile taxa and cryptobenthic and pelagic fishes were primarily influenced, positively, by seeding. As different groups of organisms responded most strongly to different types of inter- vention, eco-engineering projects aimed at maximising the biodiversity of multiple functional groups might benefit from the creation of a variety of different types of microhabitats on any given structure, that increase the breadth of niche space available to organisms [19]. Alterna- tively, projects that focus on enhancing individual species (e.g. large predators or protected species) may want to target particular interventions that have the greatest benefit for these [72, 73]. Our small-scale tile experiment tested the effects of adding habitat structure and seeding with S. glomerata on the species density and abundances of four functional groups, at two sites and through time. Using this design, it was not possible to disentangle the effects of the initial treatment conditions from the effects of later colonising biota on the responses of the four functional groups. Further research, which manipulates different size classes and numbers of

live and dead oysters, across multiple sites and locations, over which there is a greater range of species, is needed to inform the design of eco-engineering interventions. Such research should consider how site characteristics, the local species pool, and the scale of complexity relative to the size of the organisms influences outcomes of habitat manipulation [74]. When coupled with information on the costs and benefits of eco-engineering interventions, and the scales of interventions that maximise ecological benefits, the results of such studies will enable manag- ers to make informed decisions regarding the best measures, given goals and site characteristics. Supporting information S1 Table. List of functional groups and species utilising each of the four experimental treatments during the experiment. Species/morphospecies are marked as present (+) or PLOS ONE Effects of microhabitats and seeding with oysters on intertidal biodiversity PLOS ONE | https://doi.org/10.1371/journal.pone.0230807 July 16, 2020 16 / 21

absent (-) based on sampling at month 12, whereby sessile and mobile species were censused using destructive sampling and pelagic and cryptic fish were sampled using GoPros in situ. If species were recorded at months 1 and/or 6, but not at month 12, they are listed but marked as absent at 12 months. (DOCX) S2 Table. Results of generalised linear models testing the effects of habitat structure (flat vs. complex tiles) and microhabitat identity (crevice vs. ridge, nested within the complex tiles), on the number of live seeded S. glomerata sampled in-situ. The surface area of the tiles or microhabitats (offset), site and month (repeated measure) were also included in the model. Post hoc tests for significant factors of interest are shown. Tests significant at α = 0.05 are shown in bold. (DOCX) S3 Table. Results of generalised linear models testing the effects of habitat structure (flat vs. complex) and microhabitat identity (crevice vs. ridge, nested within the complex tiles) and seeding with oysters (unseeded [US] vs. seeded [S]) on the percentage cover of S. glo- merata sampled in-situ through time. The surface area of the tiles or microhabitats (offset), site and month (repeated measure) were also included in the model. Post hoc tests for signifi- cant factors of interest are shown. Tests significant at α = 0.05 are shown in bold. (DOCX) S4 Table. Results of generalised linear models testing the effects of seeding with oysters (unseeded vs. seeded) and habitat structure (flat vs. complex tiles) or microhabitat identity

pled destructively at month 12. The surface area of tiles or microhabitats (offset), site and month (repeated measure) were also included in the model. Post hoc tests for significant fac- tors of interest are shown. Tests significant at α = 0.05 are shown in bold. (DOCX) S7 Table. Results of generalised linear models testing the effects of seeding with oysters (unseeded vs. seeded) and microhabitat identity (crevice or ridge each tested separately vs. PLOS ONE Effects of microhabitats and seeding with oysters on intertidal biodiversity PLOS ONE | https://doi.org/10.1371/journal.pone.0230807 July 16, 2020 17 / 21

flat) on the environmental variables (maximum temperature, minimum temperature, standard deviation temperature, average humidity and average light). Site and month (repeated measure) were also included in the model. Post hoc tests for significant factors of interest are shown. Tests significant at α = 0.05 are shown in bold. (DOCX) Acknowledgments We thank the many people that helped in deploying and monitoring the experiment, including Emma Wilkie, Peter Simpson, Stephanie Bagala, Caleb Rankin, Dominic McAfee, Maria Vozzo and the Port Stephens Fisheries Institute and the NSW Department of Primary Industry for providing the oysters. This study was part of the World Harbour Project. Author Contributions Conceptualization: Elisabeth Marijke Anne Strain, Vivian Ruth Cumbo, Rebecca Louise Mor- ris, Peter David Steinberg, Melanie Jane Bishop. Data curation: Elisabeth Marijke Anne Strain, Vivian Ruth Cumbo, Rebecca Louise Morris, Melanie Jane Bishop. Formal analysis: Elisabeth Marijke Anne Strain. Funding acquisition: Elisabeth Marijke Anne Strain, Peter David Steinberg, Melanie Jane Bishop. Investigation: Elisabeth Marijke Anne Strain. Methodology: Elisabeth Marijke Anne Strain. Project administration: Elisabeth Marijke Anne Strain. Writing – original draft: Elisabeth Marijke Anne Strain. Writing – review & editing: Elisabeth Marijke Anne Strain, Vivian Ruth Cumbo, Rebecca Louise Morris, Peter David Steinberg, Melanie Jane Bishop. References 1. Connor EF, McCoy ED. The statistics and biology of the species-area relationship. The American Natu- ralist. 1979; 113(6):791–833.

2. Kovalenko KE, Thomaz SM, Warfe DM. Habitat complexity: approaches and future directions. Hydro- biologia. 2012; 685(1):1–17. 3. McGuinness KA. Species–area curves. Biological Reviews. 1984; 59(3):423–40. 4. Gee J, Warwick R. Metazoan community structure in relation to the fractal dimensions of marine macro- algae. Marine Ecology Progress Series. 1994; 103:141–50. 5. Matias MG, Underwood A, Coleman RA. Interactions of components of habitats alter composition and variability of assemblages. Journal of Animal Ecology. 2007; 76(5):986–94. https://doi.org/10.1111/j. 1365-2656.2007.01277.x PMID: 17714277 6. Loke LH, Todd PA. Structural complexity and component type increase intertidal biodiversity indepen- dently of area. Ecology. 2016; 97(2):383–93. https://doi.org/10.1890/15-0257.1 PMID: 27145613 7. Strain EM, Morris RL, Coleman RA, Figueira W, Steinberg P, Johnston E, et al. Increasing microhabitat complexity on seawalls can reduce fish predation on native oysters. Ecological Engineering. 2018; 120:637–42. 8. Coull BC, Wells J. Refuges from fish predation: experiments with phytal meiofauna from the New Zea- land rocky intertidal. Ecology. 1983; 64(6):1599–609. PLOS ONE Effects of microhabitats and seeding with oysters on intertidal biodiversity PLOS ONE | https://doi.org/10.1371/journal.pone.0230807 July 16, 2020 18 / 21

9. Pearsons TN, Li HW, Lamberti GA. Influence of habitat complexity on resistance to flooding and resil- ience of stream fish assemblages. Transactions of the American Fisheries Society. 1992; 121(4):427– 36. 10. McAbendroth L, Ramsay P, Foggo A, Rundle S, Bilton D. Does macrophyte fractal complexity drive invertebrate diversity, biomass and body size distributions? Oikos. 2005; 111(2):279–90. 11. Taniguchi H, Tokeshi M. Effects of habitat complexity on benthic assemblages in a variable environ- ment. Freshwater Biology. 2004; 49(9):1164–78. 12. Airoldi L, Connell SD, Beck MW. The loss of natural habitats and the addition of artificial substrata. Marine Hard Bottom Communities: Springer; 2009. p. 269–80. 13. Airoldi L, Turon X, Perkol-Finkel S, Rius M. Corridors for aliens but not for natives: effects of marine urban sprawl at a regional scale. Diversity and Distributions. 2015; 21(7):755–68. 14. Bulleri F, Chapman MG. The introduction of coastal infrastructure as a driver of change in marine envi- ronments. Journal of Applied Ecology. 2010; 47(1):26–35. 15. Chapman M. Paucity of mobile species on constructed seawalls: effects of urbanization on biodiversity. Marine Ecology Progress Series. 2003; 264:21–9. 16. Chapman M, Underwood A, Browne MA. An assessment of the current usage of ecological engineering and reconciliation ecology in managing alterations to habitats in urban estuaries. Ecological Engineer- ing. 2018; 120:560–73. 17. Chapman MG, Blockley DJ. Engineering novel habitats on urban infrastructure to increase intertidal bio-

diversity. Oecologia. 2009; 161(3):625–35. https://doi.org/10.1007/s00442-009-1393-y PMID: 19551409 18. Firth LB, Knights AM, Bridger D, Evans AJ, Mieszkowska N, Moore PJ, et al. Ocean sprawl: challenges and opportunities for biodiversity management in a changing world. Oceanography and Marine Biology: An Annual Review; 2016. p. 201–78. 19. Strain EM, Olabarria C, Mayer-Pinto M, Cumbo V, Morris RL, Bugnot AB, et al. Eco-engineering urban infrastructure for marine and coastal biodiversity: Which interventions have the greatest ecological ben- efit? Journal of Applied Ecology. 2018; 55(1):426–41. 20. Hacker SD, Steneck RS. Habitat Architecture and the Abundance and Body-Size-Dependent Habitat Selection of a Phytal Amphipod. Ecology. 1990; 71(6):2269–85. 21. Garrity SD. Some adaptations of gastropods to physical stress on a tropical rocky shore. Ecology. 1984; 65(2):559–74. 22. Underwood A, Jernakoff P. The effects of tidal height, wave-exposure, seasonality and rock-pools on grazing and the distribution of intertidal macroalgae in New South Wales. Journal of Experimental Marine Biology and Ecology. 1984; 75(1):71–96. 23. Browne MA, Chapman MG. Mitigating against the loss of species by adding artificial intertidal pools to existing seawalls. Marine Ecology Progress Series. 2014; 497:119–29. 24. Glasby T. Effects of shading on subtidal epibiotic assemblages. Journal of Experimental Marine Biology and Ecology. 1999; 234(2):275–90. 25. Irving AD, Connell SD. Sedimentation and light penetration interact to maintain heterogeneity of subtidal

habitats: algal versus invertebrate dominated assemblages. Marine Ecology Progress Series. 2002; 245:83–91. 26. Miller RJ, Etter RJ. Shading facilitates sessile invertebrate dominance in the rocky subtidal Gulf of Maine. Ecology. 2008; 89(2):452–62. https://doi.org/10.1890/06-1099.1 PMID: 18409434 27. McAfee D, Cole VJ, Bishop MJ. Latitudinal gradients in ecosystem engineering by oysters vary across habitats. Ecology. 2016; 97(4):929–39. https://doi.org/10.1890/15-0651.1 PMID: 27220209 28. Almany GR. Priority effects in coral reef fish communities. Ecology. 2003; 84(7):1920–35. 29. Dean T, Hurd L. Development in an estuarine fouling community: the influence of early colonists on later arrivals. Oecologia. 1980; 46(3):295–301. https://doi.org/10.1007/BF00346255 PMID: 28310035 30. Vieira EA, Flores AA, Dias GM. Persistence and space preemption explain species-specific founder effects on the organization of marine sessile communities. Ecology and Evolution. 2018; 8(6):3430–42. https://doi.org/10.1002/ece3.3853 PMID: 29607036 31. Connell JH, Slatyer RO. Mechanisms of succession in natural communities and their role in community stability and organization. The American Naturalist. 1977; 111(982):1119–44. 32. Harris LG, Ebeling AW, Laur DR, Rowley RJ. Community recovery after storm damage: a case of facili- tation in primary succession. Science. 1984; 224(4655):1336–8. https://doi.org/10.1126/science.224. 4655.1336 PMID: 17837192 PLOS ONE Effects of microhabitats and seeding with oysters on intertidal biodiversity

PLOS ONE | https://doi.org/10.1371/journal.pone.0230807 July 16, 2020 19 / 21

33. Palomo M, People J, Chapman M, Underwood A. Separating the effects of physical and biological aspects of mussel beds on their associated assemblages. Marine Ecology Progress Series. 2007; 344:131–42. 34. Summerhayes SA, Bishop MJ, Leigh A, Kelaher BP. Effects of oyster death and shell disarticulation on associated communities of epibiota. Journal of Experimental Marine Biology and Ecology. 2009; 379(1– 2):60–7. 35. Andre´ C, Jonsson P, Lindegarth M. Predation on settling bivalve larvae by benthic suspension feeders: the role of hydrodynamics and larval behaviour. Marine Ecology Progress Series. 1993:183–92. 36. Commito JA, Boncavage EM. Suspension-feeders and coexisting infauna: an enhancement counterex- ample. Journal of Experimental Marine Biology and Ecology. 1989; 125(1):33–42. 37. Gratwicke B, Speight MR. Effects of habitat complexity on Caribbean marine fish assemblages. Marine Ecology Progress Series. 2005; 292:301–10. 38. Morris RL, Chapman MG, Firth LB, Coleman RA. Increasing habitat complexity on seawalls: Investigat- ing large-and small-scale effects on fish assemblages. Ecology and Evolution. 2017; 7(22):9567–79. https://doi.org/10.1002/ece3.3475 PMID: 29187990 39. Ushiama S, Mayer-Pinto M, Bugnot A, Johnston E, Dafforn K. Eco-engineering increases habitat avail- ability and utilisation of seawalls by fish. Ecological Engineering. 2019; 138:403–11. 40. Ferse SCA. Artificial reef structures and coral transplantation: fish community responses and effects on

coral recruitment in North Sulawesi/Indonesia: Dissertation. University of Bremen, Bremen, Germany; 2008. 41. Gregalis KC, Johnson MW, Powers SP. Restored oyster reef location and design affect responses of resident and transient fish, crab, and shellfish species in Mobile Bay, Alabama. Transactions of the American Fisheries Society. 2009; 138(2):314–27. 42. Ng CSL, Lim SC, Ong JY, Teo LMS, Chou LM, Chua KE, et al. Enhancing the biodiversity of coastal defence structures: transplantation of nursery-reared reef biota onto intertidal seawalls. Ecological Engineering. 2015; 82:480–6. 43. Scanes E, Johnston EL, Cole VJ, O’Connor WA, Parker LM, Ross PM. Quantifying abundance and dis- tribution of native and invasive oysters in an urbanised estuary. Aquatic Invasions. 2016; 11(4):425–36. 44. Kohler KE, Gill SM. Coral Point Count with Excel extensions (CPCe): A Visual Basic program for the determination of coral and substrate coverage using random point count methodology. Computers and Geosciences. 2006; 32(9):1259–69. 45. Cappo MM, Harvey EE, Shortis MM, editors. Counting and measuring fish with baited video tech- niques-an overview. Proceedings of the Australian Society for Fish Biology Workshop, Hobart, August 2007-pages: 101–114; 2007. 46. R Core Team. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Founda- tion for Statistical Computing; 2016. 47. Wickham H. ggplot2. Computational Statistics. 2011; 3(2):180–5. 48. McAfee D O’connor WA, Bishop MJ. Fast-growing oysters show reduced capacity to provide a thermal

refuge to intertidal biodiversity at high temperatures. Journal of Animal Ecology. 2017; 86(6):1352–62. https://doi.org/10.1111/1365-2656.12757 PMID: 28913869 49. Bishop MJ, Mayer-Pinto M, Airoldi L, Firth LB, Morris RL, Loke LHL, et al. Effects of ocean sprawl on ecological connectivity: impacts and solutions. Journal of Experimental Marine Biology and Ecology. 2017. 50. Archambault P, Bourget E. Scales of coastal heterogeneity and benthic intertidal species richness, diversity and abundance. Marine Ecology Progress Series. 1996; 136(1):111–21. 51. McQuaid CD, Porri F, Nicastro K, Zardi G. Simple, scale-dependent patterns emerge from very complex effects: an example from the intertidal mussels Mytilus galloprovincialis and Perna perna. Oceanogra- phy and Marine Biology: An Annual Review. 2015; 53:127–56. 52. MacArthur M, Naylor LA, Hansom JD, Burrows MT, Loke LH, Boyd I. Maximising the ecological value of hard coastal structures using textured formliners. Journal of Ecological Engineering. 2019; 1:100002. 53. Bateman D, Bishop MJ. The environmental context and traits of habitat-forming bivalves influence the magnitude of their ecosystem engineering. Marine Ecology Progress Series. 2016; 563:95–110. 54. Grabowski JH, Peterson CH. Restoring oyster reefs to recover ecosystem services. Ecosystems engi- neers: plants to protists. 42007. p. 281–98. 55. Wilkie EM, Bishop MJ, O’Connor WA. Are native Saccostrea glomerata and invasive Crassostrea gigas oysters’ habitat equivalents for epibenthic communities in south-eastern Australia? Journal of Experi-

mental Marine Biology and Ecology. 2012; 420:16–25. PLOS ONE Effects of microhabitats and seeding with oysters on intertidal biodiversity PLOS ONE | https://doi.org/10.1371/journal.pone.0230807 July 16, 2020 20 / 21

56. Butman CA, Grassle JP, Webb CM. Substrate choices made by marine larvae settling in still water and in a flume flow. Nature. 1988; 333(6175):771. 57. Clynick B, Chapman M, Underwood A. Effects of epibiota on assemblages of fish associated with urban structures. Marine Ecology Progress Series. 2007; 332:201–10. 58. Clynick B, Chapman M, Underwood A. Fish assemblages associated with urban structures and natural reefs in Sydney, Australia. Austral Ecology. 2008; 33(2):140–50. 59. Anderson M, Connell S. Predation by fish on intertidal oysters. Marine Ecology Progress Series. 1999; 187:203–11. 60. Connell S. Predatory fish do not always affect the early development of epibiotic assemblages. Journal of Experimental Marine Biology and Ecology. 2001; 260(1):1–12. https://doi.org/10.1016/s0022-0981 (01)00246-5 PMID: 11358568 61. Connell S, Anderson M. Predation by fish on assemblages of intertidal epibiota: effects of predator size and patch size. Journal of Experimental Marine Biology and Ecology. 1999; 241(1):15–29. 62. Hedge LH, Johnston EL. Colonisation of the non-indigenous Pacific oyster Crassostrea gigas deter- mined by predation, size and initial settlement densities. PloS one. 2014; 9(3):e90621. https://doi.org/ 10.1371/journal.pone.0090621 PMID: 24663029 63. Bulleri F. Experimental evaluation of early patterns of colonisation of space on rocky shores and sea- walls. Marine Environmental Research. 2005; 60(3):355–74. https://doi.org/10.1016/j.marenvres.2004. 12.002 PMID: 15769504

64. Bulleri F. Role of recruitment in causing differences between intertidal assemblages on seawalls and rocky shores. Marine Ecology Progress Series. 2005; 287:53–65. 65. Bulleri F, Chapman M, Underwood A. Intertidal assemblages on seawalls and vertical rocky shores in Sydney Harbour, Australia. Austral Ecology. 2005; 30(6):655–67. 66. Ivesˇa L, Chapman M, Underwood AJ, Murphy RJ. Differential patterns of distribution of limpets on inter- tidal seawalls: experimental investigation of the roles of recruitment, survival and competition. Marine Ecology Progress Series. 2010; 407:55–69. 67. Jackson A, Chapman M, Underwood A. Ecological interactions in the provision of habitat by urban development: whelks and engineering by oysters on artificial seawalls. Austral Ecology. 2008; 33 (3):307–16. 68. Loke LH, Heery EC, Lai S, Bouma TJ, Todd PA. Area-independent effects of water-retaining features on intertidal biodiversity on eco-engineered seawalls in the tropics. Frontiers in Marine Science. 2019; 6:16. 69. Moreira J, Chapman M, Underwood A. Maintenance of chitons on seawalls using crevices on sand- stone blocks as habitat in Sydney Harbour, Australia. Journal of Experimental Marine Biology and Ecol- ogy. 2007; 347(1):134–43. 70. Ferrario F, Ivesˇa L, Jaklin A, Perkol-Finkel S, Airoldi L. The overlooked role of biotic factors in controlling the ecological performance of artificial marine habitats. Journal of Applied Ecology. 2016; 53(1):16–24. 71. Goff M. Evaluating habitat enhancements of an urban intertidal seawall: Ecological responses and man-

agement implications: University of Washington; 2010. 72. Hellyer CB, Harasti D, Poore AG. Manipulating artificial habitats to benefit seahorses in Sydney Har- bour, Australia. Aquatic Conservation: Marine and Freshwater Ecosystems. 2011; 21(6):582–9. 73. Toft JD, Ogston AS, Heerhartz SM, Cordell JR, Flemer EE. Ecological response and physical stability of habitat enhancements along an urban armored shoreline. Ecological Engineering. 2013; 57:97–108. 74. Mayer-Pinto M, Cole V, Johnston E, Bugnot A, Hurst H, Airoldi L, et al. Functional and structural responses to marine urbanisation. Environmental Research Letters. 2018; 13(1):014009. PLOS ONE Effects of microhabitats and seeding with oysters on intertidal biodiversity PLOS ONE | https://doi.org/10.1371/journal.pone.0230807 July 16, 2020 21 / 21