Philippa Kaur

Leclère, D. et al. Bending the curve of terrestrial biodiversity needs an integrated strategy. Nature 585, 551–556 (2020).Article 
PubMed 

Google Scholar 
Pillay, R. et al. Tropical forests are home to over half of the world’s vertebrate species. Front. Ecol. Environ. 20, 10–15 (2022).Article 
PubMed 

Google Scholar 
Turubanova, S., Potapov, P. V., Tyukavina, A. & Hansen, M. C. Ongoing primary forest loss in Brazil, Democratic Republic of the Congo, and Indonesia. Environ. Res. Lett. 13, 074028 (2018).Article 

Google Scholar 
Matricardi, E. A. T. et al. Long-term forest degradation surpasses deforestation in the Brazilian Amazon. Science 369, 1378–1382 (2020).Article 
CAS 
PubMed 

Google Scholar 
Gibson, L. et al. Primary forests are irreplaceable for sustaining tropical biodiversity. Nature 478, 378–381 (2011).Article 
CAS 
PubMed 

Google Scholar 
Barlow, J. et al. Anthropogenic disturbance in tropical forests can double biodiversity loss from deforestation. Nature 535, 144–147 (2016).Article 
CAS 
PubMed 

Google Scholar 
Watson, J. E. M. et al. The exceptional value of intact forest ecosystems. Nat. Ecol. Evol. 2, 599–610 (2018).Article 
PubMed 

Google Scholar 
Hansen, A. et al. Global humid tropics forest structural condition and forest structural integrity maps. Sci. Data 6, 232 (2019).Article 
PubMed 
PubMed Central 

Google Scholar 
Hansen, A. J. et al. A policy-driven framework for conserving the best of Earth’s remaining moist tropical forests. Nat. Ecol. Evol. 4, 1377–1384 (2020).Article 
PubMed 
PubMed Central 

Google Scholar 
COP 11 Decision X/2. Strategic Plan for Biodiversity 2011–2020 (Convention on Biological Diversity, 2010).New York Declaration on Forests (UN, 2014).Transforming our World: The 2030 Agenda for Sustainable Development. A/RES/70/1 Resolution Adopted by the United Nations General Assembly (UN, 2015).Adoption of the Paris Agreement. Proposal by the President. Draft Decision -/CP.21 (UNFCCC, 2015).Hansen, A. J. et al. Toward monitoring forest ecosystem integrity within the post-2020 Global Biodiversity Framework. Conserv. Lett. 14, e12822 (2021).Article 

Google Scholar 
Scholes, R. et al. (eds) Summary for Policymakers of the Assessment Report on Land Degradation and Restoration (IPBES, 2018).First Draft of the Post-2020 Global Biodiversity Framework (Convention on Biological Diversity, 2021).Williams, B. A. et al. Change in terrestrial human footprint drives continued loss of intact ecosystems. One Earth 3, 371–382 (2020).Article 

Google Scholar 
The IUCN Red List of Threatened Species Version 2020–1 (IUCN, 2020).Dinerstein, E. et al. An ecoregion-based approach to protecting half the terrestrial realm. Bioscience 67, 534–545 (2017).Article 
PubMed 
PubMed Central 

Google Scholar 
Ives, A. R. & Garland, T. Phylogenetic logistic regression for binary dependent variables. Syst. Biol. 59, 9–26 (2010).Article 
PubMed 

Google Scholar 
Di Marco, M., Ferrier, S., Harwood, T. D., Hoskins, A. J. & Watson, J. E. M. Wilderness areas halve the extinction risk of terrestrial biodiversity. Nature 573, 582–585 (2019).Article 
PubMed 

Google Scholar 
Betts, M. G. et al. Global forest loss disproportionately erodes biodiversity in intact landscapes. Nature 547, 441–444 (2017).Article 
CAS 
PubMed 

Google Scholar 
Fletcher, R. & Fortin, M.-J. Spatial Ecology and Conservation Modeling: Applications with R (Springer, 2018). https://doi.org/10.1007/978-3-030-01989-1Briant, G., Gond, V. & Laurance, S. G. W. Habitat fragmentation and the desiccation of forest canopies: a case study from eastern Amazonia. Biol. Conserv. 143, 2763–2769 (2010).Article 

Google Scholar 
Anderegg, W. R. L. et al. Climate-driven risks to the climate mitigation potential of forests. Science 368, eaaz7005 (2020).Article 
CAS 
PubMed 

Google Scholar 
Pillay, R. et al. Using interview surveys and multispecies occupancy models to inform vertebrate conservation. Conserv. Biol. 36, e13832 (2022).Article 
PubMed 

Google Scholar 
Agresti, A. Categorical Data Analysis (John Wiley and Sons, 2002).Smith, A. C., Koper, N., Francis, C. M. & Fahrig, L. Confronting collinearity: comparing methods for disentangling the effects of habitat loss and fragmentation. Landsc. Ecol. 24, 1271–1285 (2009).Article 

Google Scholar 
Mittermeier, R. A. et al. Wilderness and biodiversity conservation. Proc. Natl Acad. Sci. USA 18, 10309–10313 (2003).Article 

Google Scholar 
Turner, I. M. & Corlett, R. T. The conservation value of small, isolated fragments of lowland tropical rain forest. Trends Ecol. Evol. 11, 330–333 (1996).Article 
CAS 
PubMed 

Google Scholar 
Tulloch, A. I. T., Barnes, M. D., Ringma, J., Fuller, R. A. & Watson, J. E. M. Understanding the importance of small patches of habitat for conservation. J. Appl. Ecol. 53, 418–429 (2016).Article 

Google Scholar 
Wintle, B. A. et al. Global synthesis of conservation studies reveals the importance of small habitat patches for biodiversity. Proc. Natl Acad. Sci. USA 116, 909–914 (2019).Article 
CAS 
PubMed 

Google Scholar 
Hansen, M. C. et al. The fate of tropical forest fragments. Sci. Adv. 6, eaax8574 (2020).Article 
PubMed 
PubMed Central 

Google Scholar 
Prugh, L. R., Hodges, K. E., Sinclair, A. R. E. & Brashares, J. S. Effect of habitat area and isolation on fragmented animal populations. Proc. Natl Acad. Sci. USA 105, 20770–20775 (2008).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Grantham, H. S. et al. Anthropogenic modification of forests means only 40% of remaining forests have high ecosystem integrity. Nat. Commun. 11, 5978 (2020).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Beyer, H. L., Venter, O., Grantham, H. S. & Watson, J. E. M. Substantial losses in ecoregion intactness highlight urgency of globally coordinated action. Conserv. Lett. 13, e12692 (2020).Article 

Google Scholar 
Ehbrecht, M. et al. Global patterns and climatic controls of forest structural complexity. Nat. Commun. 12, 519 (2021).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
França, F. et al. Do space-for-time assessments underestimate the impacts of logging on tropical biodiversity? An Amazonian case study using dung beetles. J. Appl. Ecol. 53, 1098–1105 (2016).Article 

Google Scholar 
Di Marco, M., Venter, O., Possingham, H. P. & Watson, J. E. M. Changes in human footprint drive changes in species extinction risk. Nat. Commun. 9, 4621 (2018).Article 
PubMed 
PubMed Central 

Google Scholar 
Betts, M. G. et al. Forest degradation drives widespread avian habitat and population declines. Nat. Ecol. Evol. 6, 709–719 (2022).Article 
PubMed 
PubMed Central 

Google Scholar 
Bar-On, Y. M., Phillips, R. & Milo, R. The biomass distribution on Earth. Proc. Natl Acad. Sci. USA 115, 6506–6511 (2018).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Basset, Y. et al. Arthropod diversity in a tropical forest. Science 338, 1481–1484 (2012).Article 
CAS 
PubMed 

Google Scholar 
Cardillo, M. et al. Multiple causes of high extinction risk in large mammal species. Science 309, 1239–1241 (2005).Article 
CAS 
PubMed 

Google Scholar 
Newbold, T. et al. Ecological traits affect the response of tropical forest bird species to land-use intensity. Proc. R. Soc. B 280, 20122131 (2013).Article 
PubMed 
PubMed Central 

Google Scholar 
Maron, M., Simmonds, J. S. & Watson, J. E. M. Bold nature retention targets are essential for the global environment agenda. Nat. Ecol. Evol. 2, 1194–1195 (2018).Article 
PubMed 

Google Scholar 
Díaz, S. et al. Set ambitious goals for biodiversity and sustainability. Science 370, 411–413 (2020).Article 
PubMed 

Google Scholar 
Bird Species Distribution Maps of the World Version 2018.1 (BirdLife International, accessed 16 August 2019).Roll, U. et al. The global distribution of tetrapods reveals a need for targeted reptile conservation. Nat. Ecol. Evol. 1, 1677–1682 (2017).Article 
PubMed 

Google Scholar 
González-del-Pliego, P. et al. Phylogenetic and trait-based prediction of extinction risk for data-deficient amphibians. Curr. Biol. 29, 1557–1563 (2019).Article 
PubMed 

Google Scholar 
IUCN Habitats Classification Scheme Version 3.1 (IUCN, 2012).Böhm, M. et al. The conservation status of the world’s reptiles. Biol. Conserv. 157, 372–385 (2013).Article 

Google Scholar 
Hansen, M. C. et al. High-resolution global maps of 21st-century forest cover change. Science 342, 850–853 (2013).Article 
CAS 
PubMed 

Google Scholar 
Hansen, M. C. et al. Mapping tree height distributions in Sub-Saharan Africa using Landsat 7 and 8 data. Remote Sens. Environ. 185, 221–232 (2016).Article 

Google Scholar 
Sanderson, E. W. et al. The human footprint and the last of the wild. Bioscience 52, 891–904 (2002).Article 

Google Scholar 
Venter, O. et al. Sixteen years of change in the global terrestrial human footprint and implications for biodiversity conservation. Nat. Commun. 7, 12558 (2016).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Di Marco, M., Watson, J. E. M., Possingham, H. P. & Venter, O. Limitations and trade-offs in the use of species distribution maps for protected area planning. J. Appl. Ecol. 54, 402–411 (2017).Article 

Google Scholar 
Jenkins, C. N., Pimm, S. L. & Joppa, L. N. Global patterns of terrestrial vertebrate diversity and conservation. Proc. Natl Acad. Sci. USA 110, E2603–E2610 (2013).Article 

Google Scholar 
Simard, M., Pinto, N., Fisher, J. B. & Baccini, A. Mapping forest canopy height globally with spaceborne lidar. J. Geophys. Res. Biogeosci. 116, G04021 (2011).Article 

Google Scholar 
Sexton, J. O. et al. Global, 30-m resolution continuous fields of tree cover: Landsat-based rescaling of MODIS vegetation continuous fields with lidar-based estimates of error. Int. J. Digit. Earth 6, 427–448 (2013).Article 

Google Scholar 
Potapov, P. et al. Mapping global forest canopy height through integration of GEDI and Landsat data. Remote Sens. Environ. 253, 112165 (2021).Article 

Google Scholar 
Upham, N. S., Esselstyn, J. A. & Jetz, W. Inferring the mammal tree: species-level sets of phylogenies for questions in ecology, evolution, and conservation. PLoS Biol. 17, e3000494 (2019).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Jetz, W., Thomas, G. H., Joy, J. B., Hartmann, K. & Mooers, A. O. The global diversity of birds in space and time. Nature 491, 444–448 (2012).Article 
CAS 
PubMed 

Google Scholar 
Tonini, J. F. R., Beard, K. H., Ferreira, R. B., Jetz, W. & Pyron, R. A. Fully-sampled phylogenies of squamates reveal evolutionary patterns in threat status. Biol. Conserv. 204, 23–31 (2016).Article 

Google Scholar 
Jetz, W. & Pyron, R. A. The interplay of past diversification and evolutionary isolation with present imperilment across the amphibian tree of life. Nat. Ecol. Evol. 2, 850–858 (2018).Article 
PubMed 

Google Scholar 
Ho, L. S. T. & Ané, C. A linear-time algorithm for Gaussian and non-Gaussian trait evolution models. Syst. Biol. 63, 397–408 (2014).Article 
PubMed 

Google Scholar 
R Core Team. R: A Language and Environment for Statistical Computing (R Foundation for Statistical Computing, 2020).Verhoeven, K. J. F., Simonsen, K. L. & McIntyre, L. M. Implementing false discovery rate control: increasing your power. Oikos 108, 643–647 (2005).Article 

Google Scholar 
Benjamini, Y. & Hochberg, Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J. R. Stat. Soc. B 57, 289–300 (1995).
Google Scholar 
Bivand, R. et al. spdep: Spatial dependence: weighting schemes, statistics and models. R package version 0.7-4 (2017).Bjornstad, O. N. ncf: Spatial covariance functions. R package version 1.2-1 (2018). More

Sacristán, D., Peñarroya, B., Recatalá, L. Increasing the Knowledge on the Management of Cu-Contaminated Agricultural Soils by Cropping Tomato (Solanum Lycopersicum L.). Land Degrad. Dev. 26, 587–595 (2015).FAO. Land Degradation Assessment in Drylands. Manual for Local Level Assessment of Land Degradation and Sustainable Land Management. Part 1: Planning and Methodological Approach, Analysis and Reporting. https://www.fao.org/3/i6362e/i6362e.pdf (Food and Agriculture Organization of the United Nations, 2011).Vlachodimos, K., Papatheodorou, E. M., Diamantopoulos, J. & Monokrousos, N. Assessment of Robinia pseudoacacia cultivations as a restoration strategy for reclaimed mine spoil heaps. Environ Monit. Assess. 185, 6921–6932 (2013).Article 
CAS 
PubMed 

Google Scholar 
Misano, G. & Di Pietro, R. Habitat 9250 “Quercus trojana woods” in Italy. Fitosociologia 44, 235–238 (2007).
Google Scholar 
Biondi, E. et al. A contribution towards the knowledge of semideciduous and evergreen woods of Apulia (south-eastern Italy). Fitosociologia 41(1), 3–28 (2004).MathSciNet 

Google Scholar 
Brunetti, G. et al. Remediation of a heavy metals contaminated soil using mycorrhized and non-mycorrhized Helichrysum italicum (Roth) Don. Land Degrad. Dev. 29, 91–104 (2017).Article 

Google Scholar 
Poblador, S. et al. The influence of the invasive alien nitrogen-fixing Robinia pseudoacacia L. on soil nitrogen availability in a mixed Mediterranean riparian forest. Eur. J. For. Res. 138, 1083–1093 (2019).Article 
CAS 

Google Scholar 
Vítková, M., Müllerová, J., Sádlo, J., Pergl, J. & Pyšek, P. Black locust (Robinia pseudoacacia) beloved and despised: A story of an invasive tree in Central Europe. For. Ecol. Manag. 384, 287–302 (2017).Article 

Google Scholar 
Doran, J.W., Parkin, T.B. Quantitative indicators of soil quality: a minimum data set. in Methods for Assessing Soil Quality (eds. Doran, J.W., Jones, A.J.). 25–37 (Soil Science Society of America, 1996).Gil-Sotres, F., Trasar-Cepeda, C., Leirós, M. C. & Seoane, S. Different approaches to evaluating soil quality using biochemical properties. Soil Biol. Biochem. 37, 877–887 (2005).Article 
CAS 

Google Scholar 
Andriani, G. F. & Walsh, N. An example of the effects of anthropogenic changes on natural environment in the Apulian karst (southern Italy). Environ. Geol. 58, 313–325 (2009).Article 
ADS 

Google Scholar 
Bisantino, T., Pizzo, V., Polemio, M. & Gentile, F. Analysis of the flooding event of October 22–23, 2005 in a small basin in the province of Bari (Southern Italy). J. Agric. Eng. 531, 197–204 (2016).Article 

Google Scholar 
Soil Survey Staff. Keys to Soil Taxonomy 12th edn. (USDA-Natural Resources Conservation Service, 2014).
Google Scholar 
Tartarino, P. Inventario dei Boschi Spontanei e dei Rimboschimenti delle Provincie BAT e Bari e Stima del Loro Volume Legnoso e della sua Frazione Prelevabile nel Prossimo Ventennio. (Rapporto Tecnico Scientifico, 2011).Ismail, A. et al. Chemical composition and biological activities of Tunisian Cupressus arizonica Greene essential oils. Chem. Biodivers. 11, 150–160 (2014).Article 
CAS 
PubMed 

Google Scholar 
Navarro, A. et al. Feasibility of SRC Species for growing in Mediterranean conditions. Bioenerg. Res. 9, 208–223 (2015).Article 

Google Scholar 
Perrino, E. V., Brunetti, G. & Farrag, K. Plant communities in multi-metal contaminated soils: A case study in the National Park of Alta Murgia (Apulia Region-Southern Italy). Int. J. Phytoremediat. 16, 871–888 (2014).Article 
CAS 

Google Scholar 
VV AA Perizia Studi per il Riequilibrio Socio-Economico dell’area Interessata dall’invaso sul Torrente Locone. Consorzio Di Bonifica Apulo Lucano (1986).Lavarra, P. et al. Il Sistema Carta della Natura della Regione Puglia. (ISPRA, Serie Rapporti 204, 2014).Sparks, D. L. et al. Method of Soil Analysis: Part 3 (American Society of Agronomy Inc, 1996).Book 

Google Scholar 
Brink, R. H. Jr., Dubach, P. & Lynch, D. L. Measurement of carbohydrates in soil hydrolyzates with anthrone. Soil Sci. 89, 157–166 (1960).Article 
ADS 
CAS 

Google Scholar 
Lowry, O. H., Rosebrough, N. J., Farr, A. L. & Randall, R. J. Protein measurement with the folin phenol reagent. J. Biol. Chem. 193, 265–275 (1951).Article 
CAS 
PubMed 

Google Scholar 
García, C., Hernandez, T. & Costa, F. Potential use of dehydrogenase activity as an index of microbial activity in degraded soils. Commun. Soil Sci. Plant Anal. 28, 123–134 (1997).Article 

Google Scholar 
Vance, E. D., Brookes, P. C. & Jenkinson, D. S. An extraction method for measuring soil microbial biomass C. Soil Biol. Biochem. 19, 703–707 (1987).Article 
CAS 

Google Scholar 
Gregorich, E. G., Wen, G., Voroney, R. P. & Kachanoski, R. G. Calibration of a rapid direct chloroform extraction method for measuring soil microbial biomass C. Soil Biol. Biochem. 22, 1009–1011 (1990).Article 
CAS 

Google Scholar 
Nannipieri, P., Ceccanti, B., Cervelli, S. & Matarese, E. Extraction of phosphatase, urease, protease, organic carbon and nitrogen from soil. Soil Sci. Soc. Am. J. 44, 1011–1016 (1980).Article 
ADS 
CAS 

Google Scholar 
Tabatabai, M.A. (1994) Soil enzymes. in Methods of Soil Analysis. Part 2. Microbiological and Biochemical Properties (eds. Weaver, R.W. et al.). 775–833 (Soil Science Society of America, Inc., 1996)Traversa, A., Said-Pullicino, D., D’Orazio, V., Gigliotti, G., & Senesi, N. Properties of humic acids in Mediterranean forest soils (Southern Italy): Influence of different plant covering. Eur. J. For. Res. 130, 1045–1054 (2011)De Marco, A. et al. Decomposition of black locust and black pine leaf litter in two coeval forest stands on Mount Vesuvius and dynamics of organic components assessed through proximate analysis and NMR spectroscopy. Soil Biol. Biochem. 51, 1–15 (2012).Article 
CAS 

Google Scholar 
Wei, G. et al. Invasive Robinia pseudoacacia in China is nodulated by Mesorhizobium and Sinorhizobium species that share similar nodulation genes with native American symbionts. FEMS Microbiol. Ecol. 68, 320–328 (2009).Article 
CAS 
PubMed 

Google Scholar 
Schulze, E. D., Gebauer, G., Ziegler, H. & Lange, O. L. Estimates of nitrogen fixation by trees on an aridity gradient in Namibia. Oecologia 88, 451–455 (1991).Article 
ADS 
PubMed 

Google Scholar 
Zahran, H. H. Rhizobium-legume symbiosis and nitrogen fixation under severe conditions and in an arid climate. Microbiol. Mol. Biol. Rev. 63, 968–989 (1999).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Veste, M. & Kriebitzsch, W. U. Influence of drought stress on photosynthesis, transpiration, and growth of juvenile black locust (Robinia pseudoacacia L.). Forstarchiv 84, 35–42 (2013).
Google Scholar 
Nicolescu, V. N. et al. Ecology, growth and management of black locust (Robinia pseudoacacia L.), a non-native species integrated into European forests. J. For. Res. 31, 1081–1101 (2020).Article 
CAS 

Google Scholar 
Sposito, G. The Chemistry of Soil (Oxford University Press, 2008).
Google Scholar 
Margalef, O. et al. Global patterns of phosphatase activity in natural soils. Sci. Rep. 7, 1337. https://doi.org/10.1038/s41598-017-01418-8 (2017).Prescott, C. E. & Grayston, S. J. Tree species influence on microbial communities in litter and soil: Current knowledge and research needs. For. Ecol. Manag. 309, 19–27 (2013).Article 

Google Scholar 
Frankenberger, W. T. & Dick, W. A. Relationships between enzyme, activities and microbial growth and activity indices in soil. Soil Sci. Soc. Am. J. 47, 945–951 (1983).Article 
ADS 
CAS 

Google Scholar 
Frankenberger, W.T., Tabatabai, M.A. Amidase activity in soils III. Stability and distribution. Soil Sci. Soc. Am. J. 45, 333–338 (1981).Nannipieri, P., Trasar-Cepeda, C. & Dick, R. P. Soil enzyme activity: A brief history and biochemistry as a basis for appropriate interpretations and meta-analysis. Biol. Fertil. Soils 54, 11–19 (2018).Article 
CAS 

Google Scholar 
Pascual, J. A., Garcia, C., Hernandez, T., Moreno, J. L. & Ros, M. Soil microbial activity as a biomarker of degradation and remediation processes. Soil Biol. Biochem. 32, 1877–1883 (2000).Article 
CAS 

Google Scholar 
García-Gil, J. C., Plaza, C., Solker-Rovira, P. & Polo, A. Long-term effects of municipal solid waste compost application on soil enzyme activities and microbial biomass. Soil Biol. Biochem. 32, 1907–1913 (2000).Article 

Google Scholar 
Insam, H. & Domsch, K. H. Relationship between soil organic carbon and microbial biomass on chronosequences of reclamation sites. Microb. Ecol. 15, 177–188 (1988).Article 
CAS 
PubMed 

Google Scholar 
Acosta-Martinez, V. & Tabatabai, M. Enzyme activities in a limed agricultural soil. Biol. Fertil. Soils 31, 85–91 (2000).Article 
CAS 

Google Scholar 
Uselman, S. M., Qualls, R. G. & Thomas, R. B. A test of a potential short cut in the nitrogen cycle: the role of exudation of symbiotically fixed nitrogen from the roots of a N-fixing tree and the effects of increased atmospheric CO2 and temperature. Plant Soil 210, 21–32 (1999).Article 
CAS 

Google Scholar 
De Marco, A., Esposito, F., Berg, B., Zarrelli, A. & Virzo De Santo, A. Litter inhibitory effects on soil microbial biomass activity, and catabolic diversity in two paired stands of Robinia pseudoacacia L. and Pinus nigra Arn. Forest 9, 766. https://doi.org/10.3390/f9120766 (2018).Article 

Google Scholar 
Haghverdi, K. & Kooch, Y. Effects of diversity of tree species on nutrient cycling and soil-related processes. CATENA 178, 335–344 (2019).Article 
CAS 

Google Scholar 
Anderson, H. T. Microbial eco-physiological indicators to assess soil quality. Agric. Ecosyst. Environ. 98, 285–293 (2003).Article 

Google Scholar 
Jenkinson, D.S., Ladd, J.N. Microbial biomass in soil: Measurement and turnover. in Soil Biochemistry (eds. Paul, E.A., Ladd, J.N.). 415–471 (Marcel Dekker Inc., 1981) More

Our results show that GoM BFT larval survey samples can provide the crucial mark events for eventual CKMR estimates of adult abundance. The adult parents marked by larval samples can be directly recaptured in the fishery many years later as POPs, and indirectly through their progeny in future samples of larvae, as evidenced by the two cross-cohort HSPs (XHSPs) recovered in this study, which imply that a parent survived and spawned in the GoM in consecutive years. As more cohorts are sampled in future, the growing number of XHSPs could be used to estimate average adult survival rates, in addition to helping with the estimation of adult abundance31, as is now done for southern blue tuna40.There is a modest level of sibship within our 2016 samples, and a high level (involving over half the samples) in 2017, but it turns out not to be high enough to cause serious problems for POP-based CKMR. High sibship per se does not lead to bias in CKMR by virtue of the statistical construction of the estimate, but it does increase variance, which can be summarized through a reduction in effective sample size. In a POP-based CKMR model, our effective sample size would be about 75% of nominal for the two years combined, or 66% of nominal for the targeted sampling of 2017. Since it is actually the product of adult and juvenile sample sizes which drives precision in CKMR14, one way to think about the 75% is that we will need about 33% more adult samples to achieve a given precision on abundance estimates than if we had somehow been able to collect the same number of “independent” juvenile samples (i.e. without oversampling siblings). That increase is appreciable but entirely achievable; for WBFT, it is logistically much easier to collect more feeding-ground adult samples than to collect more larvae, and at present there is no known practical way to collect large numbers of older, more dispersed, and thus more independent, juvenile western origin bluefin tuna (WBFT).This study was motivated by the concern that sibship might be a serious impediment to use of WBFT larvae for CKMR. High levels of sibship have been found in larval collections for other taxa despite a pelagic larval phase, suggesting that abiotic factors can impede random mixing of larvae after a spawning event41. Our larval samples were only a few days old (4–11) and thus had little time to disperse since fertilization; our concern beforehand was that each tow might sample the offspring of a very small number of adults (one spawning group in one night), and in 2017 that repeatedly towing the same water mass might simply be resampling the same “family”. In practice, though, the cumulative effect was limited. Samples were not dominated by progeny from just a few adults; the maximum DPG size (i.e., number of offspring from any one adult) was 5, which is under 2% of the larval sample size. There are several possible reasons for this finding. First, plankton sample tows are typically standardized to a ten-minute duration, covering on average about 0.3 nautical miles. Based on continuous plankton cameras42, each tow is likely to tow through multiple patches of zooplankton, and therefore potentially multiple patches of BFT larvae. Second, spawning aggregations of BFT may contain many adults. For example, on the spawning grounds near the Balearic Islands in the Mediterranean, purse seine fisheries target spawning fish and individual net sets routinely capture upwards of 500 mature individuals43. These numbers suggest that BFT spawner aggregations can be quite large, although the number of individuals that contribute gametes to a single spawning event may be lower. The results of this study pose intriguing scenarios for understanding BFT larval ecology and spawning behavior, which could be explored with larger sample sizes paired with data on oceanographic conditions, direct observation of spawning aggregations, and modeling to compare observed and predicted dispersal. The results of this study are based on just two years of sampling, and numerous practical and theoretical challenges remain to fully understand BFT reproduction in the GoM.Our sibship impact calculations assume use of an unmodified adult-size-based CKMR POP model, where each juvenile is compared to each adult taking into account the latter’s size (e.g.,14). That will give unbiased estimates, which we regard as essential in a CKMR model. However, for WBFT the estimates are not fully statistically efficient, in that some adults receive more statistical weight than others because they are marked more often (by having a large DPG), and thus variance might not be the lowest achievable. Modifying the model to fix that would be simple in a “cartoon” CKMR setting where all adults are identical (e.g., Fig. 1 of14), simply by first condensing each DPG to a single representative, then only using those representatives (rather than all the larvae) in POP comparisons. Each marked parent then receives the same weight, giving maximum efficiency. For the cartoon, this condensed-DPG model still gives an unbiased estimate of abundance, because each DPG has one parent of given sex, and the chance of any sampled cartoon adult of that sex being that parent is 1/N. The DPG-condensed effective sample size is simply half the number of distinct parents, which would be a little larger than the effective sample sizes for the unmodified model shown in Table 3; e.g., in 2017, 504/2 = 252 versus 209. However, no such straightforward improvement is available for an adult-size-based CKMR model such as is needed for WABFT. Using condensed DPGs directly would bias the juvenile sampling against larger more-fecund adults, whose DPGs will tend on average to be larger and thus to experience disproportionate condensation. Those adults would be marked less often by the DPG-condensed juveniles than the model assumes, violating the basic requirements for unbiased CKMR in14. A more sophisticated model might be able to combine unbiasedness with higher efficiency but, since the unmodified adult-size-based POP model that we expect to use is unbiased and only mildly inefficient (at worst 209/252 = 83% efficient, in 2017) there seems no particular need for extra complications at present. However, that may not hold true if we eventually move to a POP + XHSP model, where the impact on unmodified CKMR variance is worse (though there is still no bias, for the same reason as with POPs). Intuitively, the biggest impact that a DPG of size 5 can have in a POP model is to suddenly raise the number of POPs by 5 if its parent happens to be sampled; within a useful total of, say, 75 POPs, the influence is not that large. But if two DPGs both of size 5 in different cohorts happen to share a parent, then the total of XHSPs suddenly jumps by 25— likely a substantial proportion of total XHSPs. Supplementary Material B also includes effective sample size formulae for a simplified XHSP-only model, which demonstrate the increased impact of within-cohort sibship; for our WBFT samples, it turns out that the XHSP-effective size is slightly lower for the targeted 2017 samples (110) than for the 2016 samples (130), unlike the POP-only effective size. Dropping from a maximum theoretical effective sample size of 252 (half the number of DPGs) down to 110 would be rather inefficient and would increase the number of years of sampling required to yield a useful XHSP dataset. This motivates developing a modified POP + XHSP model that retains unbiasedness without sacrificing too much efficiency. In principle, that can be done by condensing each DPG but then conditioning its comparison probabilities on the DPG’s original size, in accordance with the framework in14. This is a topic for subsequent research, and the results will inform future sampling strategy decisions for WBFT.One potential difficulty for western BFT CKMR might occur if a substantial proportion of animals reaching maturity are the offspring of “Western” (in genetic terms) adults who persistently spawn in the western North Atlantic but outside the GoM. However, as long as the adults marked by GoM larvae are well mixed at the time of sampling with any western adults that do spawn outside of the GoM, the total POP-based population estimate of genetically-western BFT from CKMR will remain unbiased. Given evidence from tagging of widespread adult movements within the western North Atlantic2, good mixing in the sampled feeding grounds seems likely; so, even if successful non-GoM western BFT spawning really is commonplace, there should not be a problem with relying on GoM larvae for at least the POP component of CKMR14.Studies of fish early life history have long been considered to have great potential to provide novel insight into the unique population dynamics of fishes44,45,46. Sampling efforts aimed at estimating fish recruitment dynamics have spawned a diversity of larval survey programs. Examples of these long-term programs include the California Cooperative Oceanic Fisheries Investigations, International Council for the Exploration of the Sea (ICES) surveys in the North Atlantic and adjacent areas, Southeast Monitoring and Assessment Program (SEAMAP) in the GoM, Ecosystem Monitoring (EcoMon) in the Northeast U.S., and numerous others, many of which provide indices of larval abundance widely used in fisheries and ecosystem assessments. Yet, as a result of the inherent patchiness of larvae42, sampling variability, and highly variable density dependent mortality45, fisheries scientists have often struggled to determine how larval surveys relate to the adult fish populations. Inclusion of estimates of sibship among larvae collected in surveys could refine estimates of adult spawning stock biomass estimated from these surveys.The results of this study also represent products of decades of work and coordination in obtaining high-quality DNA from larval specimens. Key steps to successful genotyping of larvae include ensuring that larvae are preserved, sorted, and handled in 95% non-denatured ethanol. In addition, strict instrument cleaning protocols must be followed, and stomachs should be removed or avoided (this study used larval tails and, when possible, eyes to avoid cross contamination of prey contents, including possible congeners and other BFT individuals). Exposure to hot lamps during the sorting and dissection processes should also be minimized to ensure that DNA quality is sufficiently high for genotyping-by-sequencing. Although the tissues available for genetic analysis were limited by the needs of other experiments that required BFT tissues, otoliths, gut contents, and other information from the same larvae, we were able to successfully genotype most larvae greater than 6 mm SL and identify thousands of informative SNPs. The lower size limit of larvae could likely be decreased if whole specimens were available for genotyping, although the use of younger larvae could increase the incidence of sibship.In summary, while we observed both FSPs and HSPs in larval collections, with elevated sibship overall and with siblings being more prevalent within tows and in nearby tows, the level of sibship was sufficiently low that collections of GoM BFT larvae can still provide the critical genetic mark of parental genotypes required for CKMR. Our results demonstrate a crucial proof of concept and are the first step towards an operational CKMR modelling estimate of spawning stock abundance for western BFT. More

Cohort demographicsA total of 206 female participants were enrolled in the study and passed our quality control standards. All participants were required to be between the ages of 18–26 years old (22.5 ± 2.1) and to be born and at the time living in one of four geographically distinct regions of the world: Barbados; Santiago, Chile; Pretoria, S. Africa; and Bangkok, Thailand. The regions do, however, differ by an order of magnitude in their geographic spread as the intra-distance separating the residence neighborhood of participants ranged from 34 (Barbados) to 681 km (Pretoria, S. Africa) (Fig. S2). The Chilean and the South African datasets are further divided into two contiguous sub-regions, or neighborhoods, to allow for a micro-geographic analysis. The study population is largely dominated by individuals with self-identified Thai heritage (33%), followed by Black African (16%), Afro-Caribbean (14%) and white (14%) descent, although 19% of the Chilean population did not report ethnicity.Study participants, despite the divergent geographies, mostly have similar dietary and lifestyle habits (Table S1). Over half the study population (62%) have a normal BMI, with the mean BMI in this range (22.6 ± 5.5). The diets of the different cohorts are also similar as of the total cohort, 78% consume a starch heavy diet (≥ 4 days a week) of rice, bread and pasta, followed by 66% who frequently consume (≥ 4 days a week) vegetables and fruit and 49% who frequently consume dairy products. The study population is split by level of tobacco exposure, with 51% of the population having never smoked, and 43% being exposed to second-hand smoke through living with a smoker. Over half (56%) of the study population own one or more pets.Stool microbiomeThe OTUs identified using the UPARSE pipeline17 were used to compute the alpha diversity of the microbial communities using the Chao1 (species richness) and Shannon (species evenness) indices. The mean Shannon indices reveal that the microbiota diversity is only significant between Thailand-Chile with FDR  More

DataWe tested our models on ten publicly available datasets. In Fig. 4 we show examples of images from each of the datasets. When applicable, the training and test splits were kept the same as in the original dataset. For example, the ZooScan, Kaggle, EILAT, and RSMAS datasets lack a specific training and test set; in these cases, benchmarks come from k-fold cross-validation51,52, and we followed the exact same procedures in order to allow for a fair comparison.Figure 4Examples of images from each of the datasets.(a) RSMAS (b) EILAT (c) ZooLake (d) WHOI (e) Kaggle (f) ZooScan (g) NA-Birds (h) Stanford dogs (i) SriLankan Beetles (j) Florida Wildtrap.Full size imageRSMAS This is a small coral dataset of 766 RGB image patches with a size of (256times 256) pixels each53. The patches were cropped out of bigger images obtained by the University of Miami’s Rosenstiel School of Marine and Atmospheric Sciences. These images were captured using various cameras in various locations. The data is separated into 14 unbalanced groups and whose labels correspond to the names of the coral species in Latin. The current SOTA for the classification of this dataset is by52. They use the ensemble of best performing 11 CNN models. The best models were chosen based on sequential forward feature selection (SFFS) approach. Since an independent test is not available, they make use of 5-fold cross-validation for benchmarking the performances.EILAT This is a coral dataset of 1123 64-pixel RGB image patches53 that were created from larger images that were taken from coral reefs near Eilat in the Red sea. The image dataset is partitioned into eight classes, with an unequal distribution of data. The names of the classes correspond to the shorter version of the scientific names of the coral species. The current SOTA52 for the classification of this dataset uses the ensemble of best performing 11 CNN models similar to RSMAS dataset and 5-fold cross-validation for benchmarking the performances.ZooLake This dataset consists of 17943 images of lake plankton from 35 classes, acquired using a Dual-magnification Scripps Plankton Camera (DSPC) in Lake Greifensee (Switzerland) between 2018 and 2020 14,54. The images are colored, with a black background and an uneven class distribution. The current SOTA22 on this dataset is based on a stacking ensemble of 6 CNN models on an independent test set.WHOI This dataset 55 contains images of marine plankton acquired by Image FlowCytobot56, from Woods Hole Harbor water. The sampling was done between late fall and early spring in 2004 and 2005. It contains 6600 greyscale images of different sizes, from 22 manually categorized plankton classes with an equal number of samples for each class. The majority of the classes belonging to phytoplankton at genus level. This dataset was later extended to include 3.4M images and 103 classes. The WHOI subset that we use was previously used for benchmarking plankton classification models51,52. The current SOTA22 on this dataset is based on average ensemble of 6 CNN models on an independent test set.Kaggle-plankton The original Kaggle-plankton dataset consists of plankton images that were acquired by In-situ Ichthyoplankton Imaging System (ISIIS) technology from May to June 2014 in the Straits of Florida. The dataset was published on Kaggle (https://www.kaggle.com/c/datasciencebowl) with images originating from the Hatfield Marine Science Center at Oregon State University. A subset of the original Kaggle-plankton dataset was published by51 to benchmark the plankton classification tasks. This subset comprises of 14,374 greyscale images from 38 classes, and the distribution among classes is not uniform, but each class has at least 100 samples. The current SOTA22 uses average ensemble of 6 CNN models and benchmarks the performance using 5-fold cross-validation.ZooScan The ZooScan dataset consists of 3771 greyscale plankton images acquired using the Zooscan technology from the Bay of Villefranche-sur-mer57. This dataset was used for benchmarking the classification models in previous plankton recognition papers51,52. The dataset consists of 20 classes with a variable number of samples for each class ranging from 28 to 427. The current SOTA22 uses average ensemble of 6 CNN models and benchmarks the performance using 2-fold cross-validation.NA-Birds NA-Birds58 is a collection of 48,000 captioned pictures of North America’s 400 most often seen bird species. For each species, there are over 100 images accessible, with distinct annotations for males, females, and juveniles, totaling 555 visual categories. The current SOTA59 called TransFG modifies the pure ViT model by adding contrastive feature learning and part selection module that replaces the original input sequence to the transformer layer with tokens corresponding to informative regions such that the distance of representations between confusing subcategories can be enlarged. They make use of an independent test set for benchmarking the model performances.Stanford Dogs The Stanford Dogs dataset comprises 20,580 color images of 120 different dog breeds from all around the globe, separated into 12,000 training images and 8,580 testing images60. The current SOTA59 makes use of modified ViT model called TransFG as explained above in NA-Birds dataset. They make use of an independent test set for benchmarking the model performances.Sri Lankan Beetles The arboreal tiger beetle data61 consists of 380 images that were taken between August 2017 and September 2020 from 22 places in Sri Lanka, including all climatic zones and provinces, as well as 14 districts. Tricondyla (3 species), Derocrania (5 species), and Neocollyris (1 species) were among the nine species discovered, with six of them being endemic . The current SOTA61 makes use of CNN-based SqueezeNet architecture and was trained using pre-trained weights of ImageNet. The benchmarking of the model performances was done on an independent test set.Florida Wild Traps The wildlife camera trap62 classification dataset comprises 104,495 images with visually similar species, varied lighting conditions, skewed class distribution, and samples of endangered species, such as Florida panthers. These were collected from two locations in Southwestern Florida. These images are categorized in to 22 classes. The current SOTA62 makes use of CNN-based ResNet-50 architecture and the performance of the model was benchmarked on an independent test set.ModelsVision transformers (ViTs)31 are an adaptation to computer vision of the Transformers, which were originally developed for natural language processing30. Their distinguishing feature is that, instead of exploiting translational symmetry, as CNNs do, they have an attention mechanism which identifies the most relevant part of an image. ViTs have recently outperformed CNNs in image classification tasks where vast amounts of training data and processing resources are available30,63. However, for the vast majority of use cases and consumers, where data and/or computational resources are limiting, ViTs are essentially untrainable, even when the network architecture is defined and no architectural optimization is required. To settle this issue, Data-efficient Image Transformers (DeiTs) were proposed32. These are transformer models that are designed to be trained with much less data and with far less computing resources32. In DeiTs, the transformer architecture has been modified to allow native distillation64, in which a student neural network learns from the results of a teacher model. Here, a CNN is used as the teacher model, and the pure vision transformer is used as the student network. All the DeiT models we report on here are DeiT-Base models32. The ViTs are ViT-B16, ViT-B32, and ViT-L32 models31.ImplementationTo train our models, we used transfer learning65: we took a model that was already pre-trained on the ImageNet43 dataset, changed the last layers depending on the number of classes, and then fine-tuned the whole network with a very low learning rate. All the models were trained with two Nvidia GTX 2080Ti GPUs.DeiTs We used DeiT-Base32 architecture, using the Python package TIMM66, which includes many of the well-known deep learning architectures, along with their pre-trained weights computed from the ImageNet dataset43. We resized the input images to 224 x 224 pixels and then, to prevent the model from overfitting at the pixel level and help it generalize better, we employed typical image augmentations during training such as horizontal and vertical flips, rotations up to 180 degrees, small zoom up’s to 20%, a small Gaussian blur, and shearing up to 10%. To handle class imbalance, we used class reweighting, which reweights errors on each example by how present that class is in the dataset67. We used sklearn utilities68 to calculate the class weights which we employed during the training phase.The training phase started with a default pytorch69 initial conditions (Kaiming uniform initializer), an AdamW optimizer with cosine annealing70, with a base learning rate of (10^{-4}), and a weight decay value of 0.03, batch size of 32 and was supervised using cross-entropy loss. We trained with early stopping, interrupting training if the validation F1-score did not improve for 5 epochs. The learning rate was then dropped by a factor of 10. We iterated until the learning rate reached its final value of (10^{-6}). This procedure amounted to around 100 epochs in total, independent of the dataset. The training time varied depending on the size of the datasets. It ranged between 20min (SriLankan Beetles) to 9h (Florida Wildtrap). We used the same procedure for all the datasets: no extra time was needed for hyperparameter tuning.ViTs We implemented the ViT-B16, ViT-B32 and ViT-L32 models using the Python package vit-keras (https://github.com/faustomorales/vit-keras), which includes pre-trained weights computed from the ImageNet43 dataset and the Tensorflow library71.First, we resized input images to 128 × 128 and employed typical image augmentations during training such as horizontal and vertical flips, rotations up to 180 degrees, small zooms up to 20%, small Gaussian blur, and shearing up to 10%. To handle class imbalance, we calculated the class weights and use them during the training phase.Using transfer learning, we imported the pre-trained model and froze all of the layers to train the model. We removed the last layer, and in its place we added a dense layer with (n_c) outputs (being (n_c) the number of classes), was preceded and followed by a dropout layer. We used the Keras-tuner72 with Bayesian optimization search73 to determine the best set of hyperparameters, which included the dropout rate, learning-rate, and dense layer parameters (10 trials and 100 epochs). After that, the model with the best hyperparameters was trained with a default tensorflow71 initial condition (Glorot uniform initializer) for 150 epochs using early stopping, which involved halting the training if the validation loss did not decrease after 50 epochs and retaining the model parameters that had the lowest validation loss.CNNs CNNs included DenseNet38, MobileNet39, EfficientNet-B240, EfficientNet-B540, EfficientNet-B640, and EfficientNet-B740 architectures. We followed the training procedure described in Ref.22, and carried out the training in tensorflow.Ensemble learningWe adopted average ensembling, which takes the confidence vectors of different learners, and produces a prediction based on the average among the confidence vectors. With this procedure, all the individual models contribute equally to the final prediction, irrespective of their validation performance. Ensembling usually results in superior overall classification metrics and model robustness74,75.Given a set of n models, with prediction vectors (vec c_i~(i=1,ldots ,n)), these are typically aggregated through an arithmetic average. The components of the ensembled confidence vector (vec c_{AA}), related to each class (alpha ) are then$$begin{aligned} c_{AA,alpha } = frac{1}{n}sum _{i=1}^n c_{i,alpha },. end{aligned}$$
(2)
Another option is to use a geometric average,$$begin{aligned} c_{GA,alpha } = root n of {prod _{i=1}^n c_{i,alpha }},. end{aligned}$$
(3)
We can normalize the vector (vec c_g), but this is not relevant, since we are interested in its largest component, (displaystyle max _alpha (c_{GA,alpha })), and normalization affects all the components in the same way. As a matter of fact, also the nth root does not change the relative magnitude of the components, so instead of (vec c_{GA}) we can use a product rule: (displaystyle max _alpha (c_{GA,alpha })=max _alpha (c_{PROD,alpha })), with (displaystyle c_{PROD,alpha } = prod _{i=1}^n c_{i,alpha }).While these two kinds of averaging are equivalent in the case of two models and two classes, they are generally different in any other case33. For example, it can easily be seen that the geometric average penalizes more strongly the classes for which at least one learner has a very low confidence value, a property that was termed veto mechanism36 (note that, while in Ref.36 the term veto is used when the confidence value is exactly zero, here we use this term in a slightly looser way). More

The evolutionary origins of zebra stripes have been investigated—and debated—for centuries. The trait is rare, conspicuous, and intensely expressed, and thus appears to beg an adaptationist explanation. However, the utility of a complete coat of densely packed, starkly contrasting black-and-white stripes is not immediately apparent. Unlike many conspicuous visual traits, striped pelage is expressed with comparable intensity in both sexes and is thus unlikely to have arisen through sexual selection alone (although in plains zebras, Equus quagga, males have stripes closer to true black than females). Stripes are clearly not aposematic warning signals, nor do they provide camouflage in either the woodland or savannah habitats common across zebra ranges1,2. So, striping presents an ideal evolutionary puzzle: a trait so refined it seems it must be “for” something, but one that confers no clear advantage upon its bearers and imposes apparent costs (conspicuousness) that cannot be explained in Zahavian terms.Scientists have proposed and investigated several possible explanations for the evolution of zebra stripes (reviewed in3). The hypotheses suggest various ways in which stripes may provide a social function (species or individual recognition or social cohesion1,4), a temperature-regulation benefit5,6, an anti-predator effect7,8, or an anti-parasite effect9,10. There is continued debate over both the merits of individual hypotheses and the likelihood of stripes having arisen via a single driver vs. a confluence or alternation of multiple selective pressures6,11.The present study addresses the hypothesis that has thus far received the most empirical support: the anti-parasite hypothesis (also known as the ectoparasite hypothesis12). Zebras, like most ungulates, are harassed by tabanid, glossinid and Stomoxys species of biting flies, which can inflict significant blood loss, transmit disease, and weaken hosts when fly-avoidance behaviors reduce the host’s feeding rate9,13,14. Yet zebras are attacked far less than sympatric ungulates across their African range15,16, and also less than other equids9,17. Zebras also produce odors that may augment their anti-fly defenses18, but so do other sympatric ungulate species18,19, and a host of observations and experiments have demonstrated that black-and-white stripes alone are unattractive, or actively repellent to tabanid, glossinid, and Stomoxys flies17,20,21,22,23.Though the effect of stripes on flies is well-established, the source of the effect remains unexplained. Since Waage’s foundational studies in the 1970s and 1980s9,24 most hypotheses have suggested ways that stripes might interfere with the visual and navigational systems of flies, making it harder for them to locate, identify, or successfully land on striped targets. These hypothetical mechanisms can be roughly grouped by the distance (and the attendant phase of a fly’s orientation and landing behavior) at which they would likely operate:

From afar: stripes might make it harder for flies to locate and distinguish zebras from background vegetation, perhaps by breaking up their outline9 or varying the way they polarize or reflect light17,31 especially from distances at which composite eyes support only low-resolution vision and cannot resolve zebra stripes as clear bands of alternating color on a single host (estimated at  > 2.0 m22,  > 4.4 m24, and even  > 20 m25).

At close range (estimates range from 0.5 to 4.0 m26): stripes might interfere with orientation or landing behavior via any of several disruptive or ‘dazzle’-related visual effects27. For example, stripes might affect ‘optic flow’, or the fly’s perceived relative motion to its target as it approaches, by creating an illusion of false direction or speed of motion (e.g., via variants of the ‘barber pole’ or ‘wagon wheel’ effects28). Alternatively, relative motion to a striped pattern within the visual field may create the perception of self-rotation, inducing the fly’s involuntary ‘optomotor response’ and resulting in an avoidance turn in an effort to stay on a straight course29.

Finally, stripes might cause confusion in the transition between long- and short-distance orientation. If zebras appear as blurred gray from a distance and then, at closer range, suddenly resolve into a sequence of floating black and white bars, this abrupt ‘visual transformation’26 might disrupt the behavioral sequence that facilitates landing.

Within these categories, hypotheses have proliferated faster than experimental tests of many of the proposed mechanisms. The very active literature on this question has grown in somewhat haphazard fashion, as curious researchers test new possibilities without eliminating old ones6. Importantly, few experiments have controlled the distance from which flies are first able to view potential landing sites (but see23). While growing evidence supports a mechanism operating at close range22,26, failing to restrict the starting distance of the fly means that the full set of possible mechanisms outlined above all remain plausible contributors to most previous results.Additionally, while many studies have, appropriately, used artificial stimuli to isolate basic effects of color, pattern, brightness, and light polarization of (usually flat) test surfaces, possible contributions of several aspects of natural zebra pelage remain untested. Controlled experiments have used various landing substrates, including striped and solid oil tray traps, sticky plastic, smooth plastic17, cloth (Experiment 2 in22), horse blankets or sheets26, and paint on live animals30. These have all clearly demonstrated a broadly replicable visual effect: stripes, and some other juxtapositions of black and white (e.g., checkerboard patterns26), repel flies. However, insofar as specific features of zebra pelage factor into proposed mechanisms of fly repellence—the reflective properties of “smooth, shiny” coats31; the orientation of the stripes17,32; the light-polarizing effects of black and white hair vs. background vegetation25; and the complex structure of hair25—there is a need for more experiments that present natural targets to wild flies (but see22,33). Similarly, most experiments have compared landing preferences between black-and-white striped, solid black, solid white, and occasionally solid grey substrates, which have served as important controls for determining that light polarization, rather than a combination of polarization and brightness, is sufficient to induce the effect of stripe avoidance17. However, it is now time to refocus on the original question by presenting flies with more realistic choices. Since biting flies seeking a bloodmeal on the African savannah seldom encounter solid black hosts, and even more rarely solid white hosts, landing choices should be compared between zebra stripes and common coat colors of sympatric mammals, namely various shades of brown. Further, tabanid, glossinid, and Stomoxys flies all avoid landing on stripes that are the same width or narrower than the widest zebra stripes 17,23, and there is some evidence that narrower stripes are even more repellent to tabanids17. This pattern is potentially significant in the application of the anti-parasite hypothesis to an adaptive explanation for the striking variation in stripe width across zebra species and between the different areas of the body on individual zebras22, but must first be confirmed with experiments using real zebra pelage.Here, we present a simple experiment designed to address each of these gaps in the literature on the anti-fly benefits of zebra stripes. In this field experiment, the landing choices of flies were tested entirely within the range at which all estimates agree flies should be able to perceive the presented stripes ( More

Angerbjörn, A. & Flux, J. E. C. Lepus timidus. Mamm. Species 1–11, https://doi.org/10.2307/3504302 (1995).Bergengren, A. On genetics, evolution and history of distribution of the heath-hare, a distinct population of the Arctic hare, Lepus timidus Lin. Swed. Wildl. (Viltrevy) 6, 381–460 (1969).
Google Scholar 
Thulin, C.-G. The distribution of mountain hares Lepus timidus in Europe: a challenge from brown hares L. europaeus? Mamm. Rev. 33, 29–42 (2003).Article 

Google Scholar 
Mills, L. S. et al. Camouflage mismatch in seasonal coat color due to decreased snow duration. Proc. Nat.Acad. Sci. 110, 7360–7365 (2013).Article 
ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Zimova, M. et al. Lack of phenological shift leads to increased camouflage mismatch in mountain hares. Proc.Royal Soc. B: Biol. Sci. 287, 20201786 (2020).Article 

Google Scholar 
Levänen, R., Kunnasranta, M. & Pohjoismäki, J. Mitochondrial DNA introgression at the northern edge of the brown hare (Lepus europaeus) range. Ann Zool Fennici 55, 15–24 (2018).Article 

Google Scholar 
Thulin, C.-G., Winiger, A., Tallian, A. G. & Kindberg, J. Hunting harvest data in Sweden indicate precipitous decline in the native mountain hare subspecies Lepus timidus sylvaticus (heath hare). J. Nat. Conserv. 64, 126069 (2021).Article 

Google Scholar 
Thulin, C.-G., Jaarola, M. & Tegelström, H. The occurrence of mountain hare mitochondrial DNA in wild brown hares. Mol. Ecol. 6, 463–467 (1997).Article 
CAS 
PubMed 

Google Scholar 
Pohjoismäki, J. L. O., Michell, C., Levänen, R. & Smith, S. Hybridization with mountain hares increases the functional allelic repertoire in brown hares. Sci. Rep. 11, 15771 (2021).Article 
ADS 
PubMed 
PubMed Central 

Google Scholar 
Hoekstra, H. E. Genetics, development and evolution of adaptive pigmentation in vertebrates. Heredity (Edinb) 97, 222–234 (2006).Article 
CAS 

Google Scholar 
Hamill, R. M., Doyle, D. & Duke, E. J. Spatial patterns of genetic diversity across European subspecies of the mountain hare, Lepus timidus L. Heredity (Edinb) 97, 355–365 (2006).Article 
CAS 

Google Scholar 
Leach, K., Montgomery, W. I. & Reid, N. Biogeography, macroecology and species’ traits mediate competitive interactions in the order Lagomorpha. Mamm. Rev. 45, 88–102 (2015).Article 

Google Scholar 
Marques, J. P. et al. Data Descriptor: Mountain hare transcriptome and diagnostic markers as resources to monitor hybridization with European hares. Sci. Data 4, 1–11 (2017).Article 

Google Scholar 
NCBI Sequence Read Archive https://identifiers.org/insdc.sra:SRP358660 (2022).Andrews, S. FastQC: a quality control tool for high throughput sequence data. Babraham Bioinformatics. Preprint at http://www.bioinformatics.babraham.ac.uk/projects/fastqc (2010).Martin, M. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet.journal 17, 10–12 (2011).Article 

Google Scholar 
Marques, J. P. et al. An annotated draft genome of the mountain hare (Lepus timidus). Genome Biol. Evol. 12, 3656–3662 (2020).Article 
CAS 
PubMed 

Google Scholar 
Li, H. & Durbin, R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics 25, 1754–1760 (2009).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Broad Institute. Picard toolkit. Broad Institute, GitHub repository. Preprint at https://broadinstitute.github.io/picard/ (2019).Garrison, E. & Marth, G. Haplotype-based variant detection from short-read sequencing. arXiv 1207.3907 (2012).Danecek, P. et al. The variant call format and VCFtools. Bioinformatics 27, 2156–2158 (2011).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Browning, S. R. & Browning, B. L. Rapid and accurate haplotype phasing and missing-data inference for whole-genome association studies by use of localized haplotype clustering. Am. J. Hum. Genet. 81, 1084–1097 (2007).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Michell, C. T., Pohjoismäki, J. L. O., Spong, G. & Thulin, C.-G. Mountain- and brown hare genetic polymorphisms to survey local adaptations and conservation status of the heath hare (Lepus timidus sylvaticus, Nilsson 1831), Dryad, https://doi.org/10.5061/dryad.3bk3j9kmp (2022).Khan, A. & Mathelier, A. Intervene: a tool for intersection and visualization of multiple gene or genomic region sets. BMC Bioinformatics 18, 287 (2017).Article 
PubMed 
PubMed Central 

Google Scholar 
R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. http://www.R-project.org/.Jombart, T. & Ahmed, I. adegenet 1.3-1: new tools for the analysis of genome-wide SNP data. Bioinformatics 27, 3070–3071 (2011).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Jombart, T. adegenet: a R package for the multivariate analysis of genetic markers. Bioinformatics 24, 1403–1405 (2008).Article 
CAS 
PubMed 

Google Scholar 
Dierckxsens, N., Mardulyn, P. & Smits, G. NOVOPlasty: De novo assembly of organelle genomes from whole genome data. Nucleic Acids Res. 45 (2017).Katoh, K. & Standley, D. M. MAFFT multiple sequence alignment software version 7: Improvements in performance and usability. Mol. Biol. Evol. 30 (2013).Trifinopoulos, J., Nguyen, L. T., von Haeseler, A. & Minh, B. Q. W-IQ-TREE: a fast online phylogenetic tool for maximum likelihood analysis. Nucleic Acids Res. 44 (2016).Kalyaanamoorthy, S., Minh, B. Q., Wong, T. K. F., von Haeseler, A. & Jermiin, L. S. ModelFinder: fast model selection for accurate phylogenetic estimates. Nat. Methods 14, 587–589 (2017).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Stamatakis, A. RaxML version 8: a tool for phylogenetic analysis and post-analysis of large phylogenies. Bioinformatics 30, 1312–1313 (2014).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Paradis, E. & Schliep, K. ape 5.0: an environment for modern phylogenetics and evolutionary analyses in R. Bioinformatics 35, 526–528 (2019).Article 
CAS 
PubMed 

Google Scholar 
Ewels, P., Magnusson, M., Lundin, S. & Käller, M. MultiQC: summarize analysis results for multiple tools and samples in a single report. Bioinformatics 32, 3047–3048 (2016).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Kamvar, Z. N., Tabima, J. F. & Grünwald, N. J. Poppr: an R package for genetic analysis of populations with clonal, partially clonal, and/or sexual reproduction. PeerJ 2, e281 (2014).Article 
PubMed 
PubMed Central 

Google Scholar 
Li, H. Minimap2: Pairwise alignment for nucleotide sequences. Bioinformatics 34, 3094–3100 (2018).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Quinlan, A. R. & Hall, I. M. BEDTools: a flexible suite of utilities for comparing genomic features. Bioinformatics 26, 841–842 (2010).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Levänen, R., Thulin, C.-G., Spong, G. & Pohjoismäki, J. L. O. Widespread introgression of mountain hare genes into Fennoscandian brown hare populations. PloS One 13, e0191790 (2018).Article 
PubMed 
PubMed Central 

Google Scholar 
Giska, I. et al. The evolutionary pathways for local adaptation in mountain hares. Mol. Ecol. 31, 1487–1503 (2022).Article 
PubMed 
PubMed Central 

Google Scholar 
Thulin, C.-G., Isaksson, M. & Tegelström, H. The origin of Scandinavian mountain hares (Lepus timidus). Gibier Faune Savage/Game and Wildlife 14, 463–475 (1997).
Google Scholar 
Ferreira, M. S. et al. The legacy of recurrent introgression during the radiation of hares. Syst. Biol. 70, 593–607 (2021).Article 
PubMed 

Google Scholar  More

Gibson, L. et al. Primary forests are irreplaceable for sustaining tropical biodiversity. Nature 478, 378–381. https://doi.org/10.1038/nature10425 (2011).Article 
ADS 
CAS 
PubMed 

Google Scholar 
Pimm, S. L. & Raven, P. Extinction by numbers. Nature 403, 843–845. https://doi.org/10.1038/35002708 (2000).Article 
ADS 
CAS 
PubMed 

Google Scholar 
​FAO. Global Forest Resources Assessment 2020: Main report. 184p (Rome, Italy, 2020).Harvey, C. A. et al. Integrating agricultural landscapes with biodiversity conservation in the Mesoamerican hotspot. Conserv Biol 22, 8–15 (2008).Article 
PubMed 

Google Scholar 
Brouwer, F. & McCarl, B. Agriculture and climate beyond 2015: A New Perspective on Future Land Use Patterns. (2006).Redo, D. J., Grau, H. R., Aide, T. M. & Clark, M. L. Asymmetric forest transition driven by the interaction of socioeconomic development and environmental heterogeneity in Central America. Proc. Natl. Acad. Sci. 109, 8839–8844. https://doi.org/10.1073/pnas.1201664109 (2012).Article 
ADS 
PubMed 
PubMed Central 

Google Scholar 
Myers, N., Mittermeier, R. A., Mittermeier, C. G., da Fonseca, G. A. B. & Kent, J. Biodiversity hotspots for conservation priorities. Nature 403, 853–858. https://doi.org/10.1038/35002501 (2000).Article 
ADS 
CAS 
PubMed 

Google Scholar 
Declerck, F. et al. Biodiversity conservation in human-modified landscapes of Mesoamerica: Past, present and future. Biol. Conserv. 143, 2301–2313. https://doi.org/10.1016/j.biocon.2010.03.026 (2010).Article 

Google Scholar 
Miller, K., Chang, E. & Johnson, N. Defining Common Ground for the Mesoamerican Biological Corridor (World Resources Institute, Washington, 2001).
Google Scholar 
Fischer, J. et al. Conservation: Limits of land sparing. Science 334, 593–593. https://doi.org/10.1126/science.334.6056.593-a (2011).Article 
ADS 
CAS 
PubMed 

Google Scholar 
Morecroft, M. D. et al. Agricultural lands key to mitigation and adaptation—Response. Science 367, 518–519. https://doi.org/10.1126/science.aba7577 (2020).Article 
ADS 
CAS 
PubMed 

Google Scholar 
Vidal, A., Kumar, C., Zinngrebe, Y., Dobie, P. & Gassner, A. Trees on farms as a nature-based solution for
biodiversity conservation in agricultural landscapes. Report number: ICRAF Policy brief No 47. 12p. World
Agroforestry Centre. https://doi.org/10.13140/RG.2.2.14852.07045 (2020).César, R. et al. Forest and landscape restoration: A review emphasizing principles, concepts, and practices. Land 10, 28. https://doi.org/10.3390/land10010028 (2020).Article 

Google Scholar 
Stanturf, J. A. et al. Implementing forest landscape restoration under the Bonn Challenge: A systematic approach. Ann. For. Sci. https://doi.org/10.1007/s13595-019-0833-z (2019).Article 

Google Scholar 
VilchezMendoza, S. et al. Consistency in bird use of tree cover across tropical agricultural landscapes. Ecol. Appl. Publ. Ecol. Soc. Am. 24, 158–168. https://doi.org/10.1890/13-0585.1 (2014).Article 

Google Scholar 
Kremen, C. & Merenlender, A. M. Landscapes that work for biodiversity and people. Science 362, eaau6020. https://doi.org/10.1126/science.aau6020 (2018).Article 
CAS 
PubMed 

Google Scholar 
Shaver, I. et al. Coupled social and ecological outcomes of agricultural intensification in Costa Rica and the future of biodiversity conservation in tropical agricultural regions. Glob. Environ. Change 32, 74–86. https://doi.org/10.1016/j.gloenvcha.2015.02.006 (2015).Article 

Google Scholar 
Zermeño-Hernández, I., Pingarroni, A. & Martínez-Ramos, M. Agricultural land-use diversity and forest regeneration potential in human- modified tropical landscapes. Agric. Ecosyst. Environ. 230, 210–220. https://doi.org/10.1016/j.agee.2016.06.007 (2016).Article 

Google Scholar 
Garibaldi, L. A. et al. Working landscapes need at least 20% native habitat. Conserv. Lett. 14, e12773. https://doi.org/10.1111/conl.12773 (2021).Article 

Google Scholar 
Estrada-Carmona, N., Martínez-Salinas, A., DeClerck, F. A. J., Vílchez-Mendoza, S. & Garbach, K. Managing the farmscape for connectivity increases conservation value for tropical bird species with different forest-dependencies. J. Environ. Manag. 250, 109504. https://doi.org/10.1016/j.jenvman.2019.109504 (2019).Article 
CAS 

Google Scholar 
Vandermeer, J. & Perfecto, I. The agroecosystem: A need for the conservation biologist’s lens. Conserv. Biol. 11, 591–592 (1997).Article 

Google Scholar 
Pardon, P. et al. Trees increase soil organic carbon and nutrient availability in temperate agroforestry systems. Agr. Ecosyst. Environ. 247, 98–111. https://doi.org/10.1016/j.agee.2017.06.018 (2017).Article 
CAS 

Google Scholar 
Nair, P. R. The coming of age of agroforestry. J. Sci. Food Agric. 87, 1613–1619. https://doi.org/10.1002/jsfa.2897 (2007).Article 
CAS 

Google Scholar 
Chatterjee, N., Nair, P. K. R., Chakraborty, S. & Nair, V. D. Changes in soil carbon stocks across the Forest-Agroforest-Agriculture/Pasture continuum in various agroecological regions: A meta-analysis. Agric. Ecosyst. Environ. 266, 55–67. https://doi.org/10.1016/j.agee.2018.07.014 (2018).Article 

Google Scholar 
Toledo-Hernández, M., Wanger, T. C. & Tscharntke, T. Neglected pollinators: Can enhanced pollination services improve cocoa yields? A review. Agr. Ecosyst. Environ. 247, 137–148. https://doi.org/10.1016/j.agee.2017.05.021 (2017).Article 

Google Scholar 
Pumariño, L. et al. Effects of agroforestry on pest, disease and weed control: A meta-analysis. Basic Appl. Ecol. 16, 573–582. https://doi.org/10.1016/j.baae.2015.08.006 (2015).Article 

Google Scholar 
Tscharntke, T. et al. Multifunctional shade-tree management in tropical agroforestry landscapes—A review. J. Appl. Ecol. 48, 619–629. https://doi.org/10.1111/j.1365-2664.2010.01939.x (2011).Article 

Google Scholar 
Martínez-Fonseca, J. G., Chávez-Velásquez, M., Williams-Guillen, K. & Chambers, C. L. Bats use live fences to move between tropical dry forest remnants. Biotropica 52, 5–10. https://doi.org/10.1111/btp.12751 (2020).Article 

Google Scholar 
Prevedello, J. A., Almeida-Gomes, M. & Lindenmayer, D. B. The importance of scattered trees for biodiversity conservation: A global meta-analysis. J. Appl. Ecol. 55, 205–214. https://doi.org/10.1111/1365-2664.12943 (2018).Article 

Google Scholar 
INE. Ministerio de Agricultura, Pesca y Alimentación (MAPA)- Gobierno de España-. 2021. Ficha de sectores. Sectores Agricultura y Pesquero. Honduras (2022).MinAmbiente-ICF. Tipologías de Bosques de Honduras. Programa ONU-REDD. Forest Carbon Partnership Facility. Tegucigalpa, Honduras. Secretaria de Energía, Recursos Naturales, Ambiente y Minas (Min Ambiente)/Instituto Nacional de Conservación y Desarrollo Forestal, Areas Protegidas y Vida Silvestre (ICF). (2017).Godinot, F., Somarriba, E., Finegan, B. & Delgado-Rodríguez, D. Secondary tropical dry forests are important to cattle ranchers in Northwestern Costa Rica. Trop. J. Environ. Sci. 54, 20–50 (2020).
Google Scholar 
Zahawi, R. A. Establishment and growth of living fence species: An overlooked tool for the restoration of degraded Areas in the Tropics. Restor. Ecol. 13, 92–102. https://doi.org/10.1111/j.1526-100X.2005.00011.x (2005).Article 

Google Scholar 
Harvey, C. A. et al. Patterns of animal diversity in different forms of tree cover in agricultural landscapes. Ecol. Appl. Publ. Ecol. Soc. Am. 16, 1986–1999. https://doi.org/10.1890/1051-0761(2006)016[1986:poadid]2.0.co;2 (2006).Article 

Google Scholar 
Miceli-Mèndez, C. L., Ferguson, B. G. & Ramìrez-Marcial, N. in Post-Agricultural Succession in the Neotropics (ed Randall W. Myster) 165–191 (Springer New York, 2008).Gaoue, O. G. & Ticktin, T. Patterns of harvesting foliage and bark from the multipurpose tree Khaya senegalensis in Benin: Variation across ecological regions and its impacts on population structure. Biol. Conserv. 137, 424–436. https://doi.org/10.1016/j.biocon.2007.02.020 (2007).Article 

Google Scholar 
Daily, G., Ceballos, G., Pacheco, J., Suzan, G. & Anchez-Azofeifa, A. Countryside biogeography of neotropical mammals: Conservation opportunities in agricultural landscapes of Costa Rica. Conserv. Biol. https://doi.org/10.1111/j.1523-1739.2003.00298.x (2003).Article 

Google Scholar 
Mayfield, M. M. & Daily, G. C. Countryside biogeography of neotropical herbaceous and shrubby plants. Ecol. Appl. 15, 423–439. https://doi.org/10.1890/03-5369 (2005).Article 

Google Scholar 
Sánchez-Merlos, D. et al. Diversidad, composición y estructura de la vegetación en un agropaisaje ganadero en Matiguás, Nicaragua. Rev. Biol. Trop. https://doi.org/10.15517/rbt.v53i3-4.14601 (2005).Article 

Google Scholar 
Sekercioglu, C. H., Loarie, S. R., Oviedo Brenes, F., Ehrlich, P. R. & Daily, G. C. Persistence of forest birds in the Costa Rican agricultural countryside. Conserv. Biol. 21, 482–494. https://doi.org/10.1111/j.1523-1739.2007.00655.x (2007).Article 
PubMed 

Google Scholar 
Wallace, G., Barborak, J. & MacFarland, C. Land use planning and regulation in and around protected areas: A study of best practices and capacity building needs in Mexico and Central America. Nat Conserv 3 (2005).
Rozendaal Danaë, M. A. et al. Biodiversity recovery of Neotropical secondary forests. Sci. Adv. 5, eaau3114. https://doi.org/10.1126/sciadv.aau3114 (2019).Article 
ADS 
PubMed 
PubMed Central 

Google Scholar 
Souza Oliveira, M. et al. Biomass of timber species in Central American secondary forests:
Towards climate change mitigation through sustainable timber harvesting. Forest Ecology and Management 496,
119439. https://doi.org/10.1016/j.foreco.2021.119439 (2021).Article 

Google Scholar 
Gillespie, T. W., Grijalva, A. & Farris, C. N. Diversity, composition, and structure of tropical dry forests in Central America. Plant Ecol. 147, 37–47. https://doi.org/10.1023/A:1009848525399 (2000).Article 

Google Scholar 
Ngo Bieng, M. A. et al. Relevance of secondary tropical forest for landscape restoration. For. Ecol. Manag. 493, 119265. https://doi.org/10.1016/j.foreco.2021.119265 (2021).Article 

Google Scholar 
Souza Oliveira, M. et al. Biomass of timber species in Central American secondary forests: Towards climate change mitigation through sustainable timber harvesting. For. Ecol. Manag. 496, 119439. https://doi.org/10.1016/j.foreco.2021.119439 (2021).Article 

Google Scholar 
Chacón, L. M. & Harvey, C. A. Live fences and landscape connectivity in a neotropical agricultural landscape. Agrofor. Syst. 68, 15. https://doi.org/10.1007/s10457-005-5831-5 (2006).Article 

Google Scholar 
Harvey, C. A. et al. Conservation value of dispersed tree cover threatened by pasture management. For. Ecol. Manag. 261, 1664–1674. https://doi.org/10.1016/j.foreco.2010.11.004 (2011).Article 

Google Scholar 
Suding, K. N. Toward an Era of restoration in ecology: Successes, failures, and opportunities ahead. Annu. Rev. Ecol. Evol. Syst. 42, 465–487. https://doi.org/10.1146/annurev-ecolsys-102710-145115 (2011).Article 

Google Scholar 
Moguel, P. & Toledo, V. M. Biodiversity conservation in traditional coffee systems of Mexico. Conserv. Biol. 13, 11–21. https://doi.org/10.1046/j.1523-1739.1999.97153.x (1999).Article 

Google Scholar 
Harrison, R. D., Harrison, S., Laumonier, Y., Somarriba, E. & Suber, M. Biodiversity monitoring for agricultural landscapes. A protocol using biodiversity metrics to monitor agricultural sustainability under Aichi Target 7. (2019).Heck, K. L. Jr., van Belle, G. & Simberloff, D. Explicit calculation of the rarefaction diversity measurement and the determination of sufficient sample size. Ecology 56, 1459–1461. https://doi.org/10.2307/1934716 (1975).Article 

Google Scholar 
Magurran, A. E. Measuring Biological Diversity (Wiley-Blackwell, New Jersey, 2004).
Google Scholar 
Chao, A. et al. Rarefaction and extrapolation with Hill numbers: a framework for sampling and estimation in species diversity studies. Ecol. Monogr. 84, 45–67. https://doi.org/10.1890/13-0133.1 (2014).Article 

Google Scholar 
Jost, L. Partitioning diversity into independent alpha and beta components. Ecology 88, 2427–2439. https://doi.org/10.1890/06-1736.1 (2007).Article 
PubMed 

Google Scholar 
Gotelli, N. J. & Colwell, R. K. Quantifying biodiversity: Procedures and pitfalls in the measurement and comparison of species richness. Ecol. Lett. 4, 379–391. https://doi.org/10.1046/j.1461-0248.2001.00230.x (2001).Article 

Google Scholar 
Zuur, A. F., Ieno, E. N., Walker, N., Saveliev, A. A. & Smith, G. M. Mixed Effects Models and Extensions in Ecology with R. XXII, 574 (Springer New York, NY, 2009).R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. (2021).Oksanen, J. et al. Vegan: Community Ecology Package. R Package Version 2.2-1 2, 1–2 (2015).Hsieh, T. C., Ma, K. & Chao, A. iNEXT: An R package for rarefaction and extrapolation of species diversity (Hill numbers). Methods Ecol. Evol. https://doi.org/10.1111/2041-210X.12613 (2016).Article 

Google Scholar 
Venables, W. N & Ripley, B. D Modern Applied Statistics with S. Fourth Edition. Springer, New York. ISBN 0-387-
95457-0 (2002)Wickham, H. ggplot2: Elegant graphics for data analysis (Springer, 2009).Book 
MATH 

Google Scholar 
gridExtra: Miscellaneous Functions for “Grid” Graphics. R package version 2.3. (2017). More