Ecology

It’s surprisingly hard to catch a sloth. Although they’re slow — very, very slow — if you climb a tree to catch one, it will move along to the next tree. Once you climb the new tree, it will move back again.My team does this regularly, as we conduct the Sloth Backpack Project, a data-logging initiative here in Costa Rica, where many sloths coexist with people. In 2017, I wanted to do more than research, so I started the Sloth Conservation Foundation.In this photograph, I’m fitting a backpack to a brown-throated three-fingered sloth (Bradypus variegatus) that we named Baguette, after a nearby bakery. The backpack will collect data on her location, movement and living patterns.We had found Baguette about 20 minutes earlier, balancing atop construction fencing as she attempted to escape two pit bulls. Baguette wasn’t all that grateful. She’s a feisty old girl. She’s old: she’s missing fingers, and she’s got scars on her face.I adore sloths, but I also envy them. They’re a powerful symbol of the slowness that our society needs more of. They don’t let anything stress them out unless it’s really important — they just get on with life.The backpack project will help us to understand sloth behaviour, so we can better protect them as the urban environment grows. This year, I received a €50,000 (US$52,220) Future For Nature award, which we will use to train a dog to detect sloth faeces. We can use faeces as a proxy for sloth numbers and locations in the region, and ultimately work out the boundaries of the species, how fast populations are declining and which conservation measures work.I’m happy I’ve moved away from academia — I can put all my energy into conservation as opposed to bashing out papers. That’s what I feel ecology should focus on — how we can use what we’re learning to give back to other species. More

Hughes, T. P. et al. Climate change, human impacts, and the resilience of coral reefs. Science 301(5635), 929–933 (2003).CAS 
PubMed 
Article 
ADS 

Google Scholar 
Hoegh-Guldberg, O. E. et al. Coral reefs under rapid climate change and ocean acidification. Science 318(5857), 1737–1742 (2007).CAS 
PubMed 
Article 
ADS 

Google Scholar 
Soares, M. et al. The flourishing and vulnerabilities of zoantharians on Southwestern Atlantic reefs. Mar. Environ. Res. 173(3), 105535 (2021).Ban, N. C. et al. Designing, implementing and managing marine protected areas: Emerging trends and opportunities for coral reef nations. J. Exp. Mar. Biol. Ecol. 408(1–2), 21–31 (2011).Article 

Google Scholar 
Magris, R. A., Pressey, R. L., Mills, M., Vila-Nova, D. A. & Floeter, S. Integrated conservation planning for coral reefs: Designing conservation zones for multiple conservation objectives in spatial prioritisation. Glob. Ecol. Conserv. 11, 53–68 (2017).Article 

Google Scholar 
Vercammen, A. et al. Evaluating the impact of accounting for coral cover in large-scale marine conservation prioritizations. Divers. Distrib. 25(10), 1564–1574 (2019).Article 

Google Scholar 
Giakoumi, S., Grantham, H. S., Kokkoris, G. D. & Possingham, H. P. Designing a network of marine reserves in the Mediterranean Sea with limited socio-economic data. Biol. Conserv. 144(2), 753–763 (2011).Article 

Google Scholar 
Gill, D. A. et al. Capacity shortfalls hinder the performance of marine protected areas globally. Nature 543(7647), 665–669 (2017).CAS 
PubMed 
Article 
ADS 

Google Scholar 
Magris, R. A. et al. A blueprint for securing Brazil’s marine biodiversity and supporting the achievement of global conservation goals. Divers. Distrib. 27(2), 198–215 (2021).Article 

Google Scholar 
Day, J. C. Zoning—lessons from the Great Barrier Reef marine park. Ocean Coast. Manag. 45(2–3), 139–156 (2002).Article 

Google Scholar 
Agardy, T. Ocean Zoning: Making Marine Management More Effective (Earthscan, 2010).Makino, A., Klein, C. J., Beger, M., Jupiter, S. D. & Possingham, H. P. Incorporating conservation zone effectiveness for protecting biodiversity in marine planning. PLoS ONE 8(11), e78986 (2013).CAS 
PubMed 
PubMed Central 
Article 
ADS 

Google Scholar 
Villa, F., Tunesi, L. & Agardy, T. Zoning marine protected areas through spatial multiple-criteria analysis: The case of the Asinara Island National Marine Reserve of Italy. Conserv. Biol. 16(2), 515–526 (2002).Article 

Google Scholar 
Muhl, E. K., Esteves Dias, A. C. & Armitage, D. Experiences with governance in three marine conservation zoning initiatives: Parameters for assessment and pathways forward. Front. Mar. Sci. 7, 629 (2020).Article 

Google Scholar 
Beger, M. et al. Integrating regional conservation priorities for multiple objectives into national policy. Nat. Commun. 6(1), 1–8 (2015).Article 
CAS 

Google Scholar 
Ban, N. C. et al. A social–ecological approach to conservation planning: Embedding social considerations. Front. Ecol. Environ. 11(4), 194–202 (2013).Article 

Google Scholar 
Teh, L. C., Teh, L. S. & Jumin, R. Combining human preference and biodiversity priorities for marine protected area site selection in Sabah, Malaysia. Biol. Conserv. 167, 396–404 (2013).Article 

Google Scholar 
Sarker, S., Rahman, M. M., Yadav, A. K. & Islam, M. M. Zoning of marine protected areas for biodiversity conservation in Bangladesh through socio-spatial data. Ocean Coast. Manag. 173, 114–122 (2019).Article 

Google Scholar 
Day, J. C., Kenchington, R. A., Tanzer, J. M. & Cameron, D. S. Marine zoning revisited: How decades of zoning the Great Barrier Reef has evolved as an effective spatial planning approach for marine ecosystem-based management. Aquat. Conserv. Mar. Freshw. Ecosyst. 29, 9–32 (2019).Article 

Google Scholar 
Claudet, J. et al. Assessing the effects of marine protected area (MPA) on a reef fish assemblage in a northwestern Mediterranean marine reserve: Identifying community-based indicators. Biol. Conserv. 130(3), 349–369 (2006).Article 

Google Scholar 
Emslie, M. J. et al. Expectations and outcomes of reserve network performance following re-zoning of the Great Barrier Reef Marine Park. Curr. Biol. 25(8), 983–992 (2015).CAS 
PubMed 
Article 

Google Scholar 
McClure, E. C. et al. Higher fish biomass inside than outside marine protected areas despite typhoon impacts in a complex reefscape. Biol. Cons. 241, 108354 (2020).Article 

Google Scholar 
Bender, M. G. et al. Local ecological knowledge and scientific data reveal overexploitation by multigear artisanal fisheries in the Southwestern Atlantic. PLoS ONE 9(10), e110332 (2014).PubMed 
PubMed Central 
Article 
ADS 
CAS 

Google Scholar 
Hamilton, R. J. et al. Hyperstability masks declines in bumphead parrotfish (Bolbometopon muricatum) populations. Coral Reefs 35(3), 751–763 (2016).Article 
ADS 

Google Scholar 
Pereira, P. H. C., Ternes, M. L. F., Nunes, J. A. C. & Giglio, V. J. Overexploitation and behavioral changes of the largest South Atlantic parrotfish (Scarus trispinosus): Evidence from fishers’ knowledge. Biol. Conserv. 254, 108940 (2021).Article 

Google Scholar 
Mumby, P. J. et al. Fishing, trophic cascades, and the process of grazing on coral reefs. Science 311(5757), 98–101 (2006).CAS 
PubMed 
Article 
ADS 

Google Scholar 
Mumby, P. J. & Harborne, A. R. Marine reserves enhance the recovery of corals on Caribbean reefs. PLoS ONE 5(1), e8657 (2010).PubMed 
PubMed Central 
Article 
ADS 
CAS 

Google Scholar 
Topor, Z. M., Rasher, D. B., Duffy, J. E. & Brandl, S. J. Marine protected areas enhance coral reef functioning by promoting fish biodiversity. Conserv. Lett. 12(4), e12638 (2019).Article 

Google Scholar 
Liu, C., White, M. & Newell, G. Measuring and comparing the accuracy of species distribution models with presence–absence data. Ecography 34(2), 232–243 (2011).CAS 
Article 

Google Scholar 
Miranda, R. J. et al. Integrating long term ecological research (LTER) and marine protected area management: Challenges and solutions. Oecol. Aust. 24(2), 279–300 (2020).Article 

Google Scholar 
ICMBIO. Plano de Manejo da Área de Proteção Ambiental Costa dos Corais. ICMBio/MMA (2021).Jones, K. R. et al. Area requirements to safeguard Earth’s marine species. One Earth 2(2), 188–196 (2020).Article 
ADS 

Google Scholar 
Figueiredo, M. S. & Grelle, C. E. V. Predicting global abundance of a threatened species from its occurrence: Implications for conservation planning. Divers. Distrib. 15(1), 117–121 (2009).Article 

Google Scholar 
Pearce, J. & Ferrier, S. The practical value of modelling relative abundance of species for regional conservation planning: A case study. Biol. Conserv. 98(1), 33–43 (2001).Article 

Google Scholar 
Ferreira, H. M., Magris, R. A., Floeter, S. R. & Ferreira, C. E. Drivers of ecological effectiveness of marine protected areas: A meta-analytic approach from the Southwestern Atlantic Ocean (Brazil). J. Environ. Manag. 301, 113889 (2021).Article 

Google Scholar 
Mills, M. et al. Real-world progress in overcoming the challenges of adaptive spatial planning in marine protected areas. Biol. Conserv. 181, 54–63 (2015).Article 

Google Scholar 
Bennett, N. J. et al. Local support for conservation is associated with perceptions of good governance, social impacts, and ecological effectiveness. Conserv. Lett. 12(4), e12640 (2019).Article 

Google Scholar 
Oldekop, J. A., Holmes, G., Harris, W. E. & Evans, K. L. A global assessment of the social and conservation outcomes of protected areas. Conserv. Biol. 30(1), 133–141 (2016).CAS 
PubMed 
Article 

Google Scholar 
Emslie, M. J. et al. Decades of monitoring have informed the stewardship and ecological understanding of Australia’s Great Barrier Reef. Biol. Conserv. 252, 108854 (2020).Article 

Google Scholar 
Gerhardinger, L. C., Godoy, E. A., Jones, P. J., Sales, G. & Ferreira, B. P. Marine protected dramas: The flaws of the Brazilian national system of marine protected areas. Environ. Manag. 47(4), 630–643 (2011).Article 
ADS 

Google Scholar 
Oliveira, E. A., Martelli, H., Silva, A. C. S. E., Martelli, D. R. B. & Oliveira, M. C. L. Science funding crisis in Brazil and COVID-19: Deleterious impact on scientific output. Anais Acad. Bras. Ciênc. 92, 1–2 (2020).
Floeter, S. R., Halpern, B. S. & Ferreira, C. E. L. Effects of fishing and protection on Brazilian reef fishes. Biol. Conserv. 128(3), 391–402 (2006).Article 

Google Scholar 
Bender, M. G., Floeter, S. R. & Hanazaki, N. Do traditional fishers recognise reef fish species declines? Shifting environmental baselines in E astern B razil. Fish. Manag. Ecol. 20(1), 58–67 (2013).Article 

Google Scholar 
Hoey, A. S. & Bonaldo, R. M. (eds) Biology of Parrotfishes (CRC Press, Boca Raton, 2018).
Google Scholar 
Frédou, T. & Ferreira, B. P. Bathymetric trends of Northeastern Brazilian snappers (Pisces, Lutjanidae): Implications for the reef fishery dynamic. Braz. Arch. Biol. Technol. 48(5), 787–800 (2005).Article 

Google Scholar 
Guerra, A. S. Wolves of the Sea: Managing human-wildlife conflict in an increasingly tense ocean. Mar. Policy 99, 369–373 (2019).Article 

Google Scholar 
Hawkins, J. P. & Roberts, C. M. Effects of fishing on sex-changing Caribbean parrotfishes. Biol. Cons. 115(2), 213–226 (2004).Article 

Google Scholar 
Tuya, F. et al. Effect of fishing pressure on the spatio-temporal variability of the parrotfish, Sparisoma cretense (Pisces: Scaridae), across the Canarian Archipelago (eastern Atlantic). Fish. Res. 7(1), 24–33 (2006).Article 

Google Scholar 
Steneck, R. S., Arnold, S. N. & Mumby, P. J. Experiment mimics fishing on parrotfish: Insights on coral reef recovery and alternative attractors. Mar. Ecol. Prog. Ser. 506, 115–127 (2014).Article 
ADS 

Google Scholar 
Taylor, B. M., Trip, E. D., & Choat, J. H. Dynamic demography: Investigations of life-history variation in the parrotfishes. In Biology of Parrotfishes 69–98 (CRC Press, 2018).Moura, R. L. & Francini-Filho, R. B. Reef and Shore Fishes of the Abrolhos Region, Brazil Vol. 38, 40–55 (RAP Bulletin of Biological Assessment, Washington, 2005).
Google Scholar 
Francini-Filho, R. B., Moura, R. L., Ferreira, C. M. & Coni, E. O. Live coral predation by parrotfishes (Perciformes: Scaridae) in the Abrolhos Bank, eastern Brazil, with comments on the classification of species into functional groups. Neotrop. Ichthyol. 6, 191–200 (2008).Article 

Google Scholar 
Freitas, M. O. et al. Age, growth, reproduction and management of Southwestern Atlantic’s largest and endangered herbivorous reef fish, Scarus trispinosus Valenciennes, 1840. PeerJ 7, e7459 (2019).PubMed 
PubMed Central 
Article 

Google Scholar 
Pinheiro, H. T. et al. An inverted management strategy for the fishery of endangered marine species. Front. Mar. Sci. 8, 172 (2021).Article 

Google Scholar 
Correia, M. D. Scleractinian corals (Cnidaria: Anthozoa) from reef ecosystems on the Alagoas coast, Brazil. J. Mar. Biol. Assoc. U. K. 91, 659–668 (2011).CAS 
Article 

Google Scholar 
Santos, D. K. F., Rufino, R. D., Luna, J. M., Santos, V. A. & Sarubbo, L. A. Biosurfactants: Multifunctional biomolecules of the 21st century. Int. J. Mol. Sci. 17(3), 401 (2016).PubMed 
PubMed Central 
Article 
CAS 

Google Scholar 
de Oliveira, S. et al. Oil spill in South Atlantic (Brazil): Environmental and governmental disaster. Mar. Policy 115, 103879 (2020).Article 

Google Scholar 
Teixeira, L. M. P. & Creed, J. C. A decade on: An updated assessment of the status of marine non-indigenous species in Brazil. Aquat. Invasions 15(1), 30–43 (2020).Article 

Google Scholar 
Braga, M. D. A. et al. Retirement risks: Invasive coral on old oil platform on the Brazilian equatorial continental shelf. Mar. Pollut. Bull. 165, 112156 (2021).CAS 
PubMed 
Article 

Google Scholar 
Luiz, O. J. et al. Multiple lionfish (Pterois spp.) new occurrences along the Brazilian coast confirm the invasion pathway into the Southwestern Atlantic. Biol. Invasions 23, 3013–3019 (2021).Article 

Google Scholar 
Maida, M., & Ferreira, B. P. Coral reefs of Brazil: An overview. In Proceedings of the 8th International Coral Reef Symposium, Vol. 1, 263–274 (Smithsonian Tropical Research Institute Panamá, 1997).Pereira, P. H. C., Macedo, C. H., Nunes, J. D. A. C., Marangoni, L. F. D. B. & Bianchini, A. Effects of depth on reef fish communities: Insights of a “deep refuge hypothesis” from Southwestern Atlantic reefs. PLoS ONE 13(9), e0203072 (2018).PubMed 
PubMed Central 
Article 
CAS 

Google Scholar 
ICMBIO. Plano de Manejo da Área de Proteção Ambiental Costa dos Corais (ICMBio/MMA, 2013).Hill, J. & Wilkinson, C. E. Methods for Ecological Monitoring of Coral Reefs Vol. 117 (Australian Institute of Marine Science, Townsville, 2004).
Google Scholar 
Dalapicolla, J. Tutorial de modelos de distribuição de espécies: guia prático usando o MaxEnt e o ArcGIS 10. Laboratório de Mastozoologia e Biogeografia. Universidade Federal do Espírito Santo, Vitória. Retrieved, 6 (2016).Phillips, S. J., Dudík, M., & Schapire, R. E. A maximum entropy approach to species distribution modeling. In Proceedings of the Twenty-First International Conference on Machine learning, Vol. 83 (2004).Phillips, S. J., Anderson, R. P. & Schapire, R. E. Maximum entropy modeling of species geographic distributions. Ecol. Model. 190(3–4), 231–259 (2006).Article 

Google Scholar 
Anderson, R. P. & Martınez-Meyer, E. Modeling species’ geographic distributions for preliminary conservation assessments: An implementation with the spiny pocket mice (Heteromys) of Ecuador. Biol. Conserv. 116(2), 167–179 (2004).Article 

Google Scholar 
Phillips, S. J., Anderson, R. P., Dudík, M., Schapire, R. E. & Blair, M. E. Opening the black box: An open-source release of Maxent. Ecography 40(7), 887–893 (2017).Article 

Google Scholar 
Rodrigues, E. D. C., Rodrigues, F. A., Rocha, R. L. A. & Corrêa, P. L. P. An adaptive maximum entropy approach for modeling of species distribution. Mem. WTA 108–117 (2010).Rodrigues, E. S. D. C., Rodrigues, F. A., Ricardo, L. D. A., Corrêa, P. L. & Giannini, T. C. Evaluation of different aspects of maximum entropy for niche-based modeling. Procedia Environ. Sci. 2, 990–1001 (2010).Article 

Google Scholar 
Hattab, T. et al. The use of a predictive habitat model and a fuzzy logic approach for marine management and planning. PLoS ONE 8(10), e76430 (2013).CAS 
PubMed 
PubMed Central 
Article 
ADS 

Google Scholar 
Galante, P. J. et al. The challenge of modeling niches and distributions for data-poor species: A comprehensive approach to model complexity. Ecography 41(5), 726–736 (2018).Article 

Google Scholar 
Silber, G. K. et al. Projecting marine mammal distribution in a changing climate. Front. Mar. Sci. 4, 413 (2017).Article 

Google Scholar 
Perkins-Taylor, I. E. & Frey, J. K. Predicting the distribution of a rare chipmunk (Neotamias quadrivittatus oscuraensis): Comparing MaxEnt and occupancy models. J. Mammal. 101(4), 1035–1048 (2020).PubMed 
PubMed Central 
Article 

Google Scholar 
Lee, C. M., Lee, D. S., Kwon, T. S., Athar, M. & Park, Y. S. Predicting the global distribution of Solenopsis geminata (Hymenoptera: Formicidae) under climate change using the MaxEnt model. Insects 12(3), 229 (2021).PubMed 
PubMed Central 
Article 

Google Scholar 
Possingham, H., Ball, I. & Andelman, S. Mathematical methods for identifying representative reserve networks. In Quantitative methods for conservation biology 291–306 (Springer, New York, 2000).Terrell, G. R. & Scott, D. W. Variable kernel density estimation.  Ann. Stat. 20(3), 1236–1265 (1992).
O’Brien, S. H., Webb, A., Brewer, M. J. & Reid, J. B. Use of kernel density estimation and maximum curvature to set Marine Protected Area boundaries: Identifying a Special Protection Area for wintering red-throated divers in the UK. Biol. Conserv. 156, 15–21 (2012).Article 

Google Scholar 
Fielding, A. H. & Bell, J. F. A review of methods for the assessment of prediction errors in conservation presence/absence models. Environ. Conserv. 24, 38–49 (1997).Article 

Google Scholar  More

Historically, species distributions have been modeled independently from each other due to unavailability of multispecies datasets and computational restraints. However, ecological datasets that provide insights about collections of organisms have become prevalent over the last decade thanks to efforts like Long Term Ecological Research Network (LTER), National Ecological Observatory Network (NEON), and citizen science surveys1. In addition, technology has improved our ability to fit modern statistical models to these datasets that account for both species environmental preferences and interspecies dependence. These advancements have allowed for the development of joint species distribution models (JSDM)2,3,4 that can model dependence among species simultaneously with environmental drivers of occurrence and/or abundance.Species distributions are shaped by both interspecies dynamics and environmental preferences5,6,7,8. JSDMs integrate both sources of variability and adjust uncertainty to reflect that multiple confounded factors can contribute to similar patterns in species distributions. Some have proposed that JSDMs not only account for biotic interactions but also correct estimates of association between species distributions and environmental drivers3,9, while others claim JSDMs cannot disentangle the roles of interspecies dependence and environmental drivers5. We address why JSDMs can provide inference distinct from their concomitant independent SDMs, how certain parameterizations of a JSDM induce confounding between the environmental and random species effects, and when deconfounding these effects may be appealing for computation and interpretation.Because of the prevalence of occupancy data for biomonitoring in ecology, we focus our discussion of community confounding in JSDMs on occupancy models, although we also consider a JSDM for species density data in the simulation study. The individual species occupancy model was first formulated by MacKenzie et al.10 and has several joint species extensions4,11,12,13,14,15,16. We chose to investigate the impacts of community confounding on the probit model since it has been widely used in the analysis of occupancy data4,13,17. We also developed a joint species extension to the Royle-Nichols model18 and consider community confounding in that model.We use the probit and Royle-Nichols occupancy models to improve our understanding of montaine mammal communities in what follows. We show that including unstructured random species effects in either occupancy model induces confounding between the fixed environmental and random species effects. We demonstrate how to orthogonalize these effects in the model and compare the resulting inference compared to models where species are treated independently.Unlike previous approaches that have applied restricted regression techniques similar to ours, we use it in the context of well-known ecological models for species occupancy and intensity. While such approaches have been discussed in spatial statistics and environmental science, they have not been adopted in settings involving the multivariate analysis of community data. We draw parallels between restricted spatial regression and restricted JSDMs but also highlight where the methods differ in goals and outcomes. We find that the computational benefits conferred by performing restricted spatial regression also hold for some joint species distribution models.Royle-Nichols joint species distribution modelWe present a JSDM extension to the Royle-Nichols model18. The Royle-Nichols model accounts for heterogeneity in detection induced by the species’ latent intensity, a surrogate related to true species abundance. Abundance, density, and occupancy estimation often requires an explicit spatial region that is closed to emmigration and immigration. In our model, the unobservable intensity variable helps us explain heterogeneity in the frequencies we observe a species at different sites without making assumptions about population closure. In the “Model” section, we further discuss the distinctions between abundance and intensity in the Royle-Nichols model.The Royle-Nichols model utilizes occupancy survey data but provides inference distinct from the basic occupancy model10. In the Royle-Nichols model, we estimate individual detection probability for homogeneous members of the population, whereas in an occupancy model, we estimate probability of observing at least one member of the population given that the site is occupied. Furthermore, the Royle-Nichols model allows us to relate environmental covariates to the latent intensity associated with a species at a site, while in an occupancy model, environmental covariates are associated with the species latent probability of occupancy at a site. Species intensity and occupancy may be governed by different mechanisms, and inference from an intensity model can be distinct from that provided by an occupancy model19,20,21. Cingolani et al.20 proposed that, in plant communities, certain environmental filters preclude species from occupying a site and an additional set of filters may regulate if a species can flourish. Hence, certain covariates that were unimportant in an occupancy model may improve predictive power in an intensity model.Community confoundingSpecies distributions are shaped by environment as well as competition and mutualism within the community8,22,23. Community confounding occurs when species distributions are explained by a convolution of environmental and interspecies effects and can lead to inferential differences between a joint and single species distribution model as well as create difficulties for fitting JSDMs. Former studies have incorporated interspecies dependence into an occupancy model4,11,12,13,14,15,16, and others have addressed spatial confounding1,17,24,25, but none of these explicitly addressed community confounding. However, all Bayesian joint occupancy models naturally attenuate the effects of community confounding due to the prior on the regression coefficients. The prior, assuming it is proper, induces regularization on the regression coefficients26 that can lessen the inferential and computational impacts of confounding27. Furthermore, latent factor models like that described by Tobler et al.4 restrict the dimensionality of the random species effect which should also reduce confounding with the environmental effects.We address community confounding by formulating a version of our model that orthogonalizes the environmental effects and random species effects. Orthogonalizing the fixed and random effects is common practice in spatial statistics and often referred to as restricted spatial regression27,28,29,30,31. Restricted regression has been applied to spatial generalized linear mixed models (SGLMM) for observations (varvec{y},) which can be expressed as$$begin{aligned} varvec{y}&sim [varvec{y}|varvec{mu }, varvec{psi }], end{aligned}$$
(1)
$$begin{aligned} g(varvec{mu })&= varvec{X}varvec{beta } + varvec{eta }, end{aligned}$$
(2)
$$begin{aligned} varvec{eta }&sim mathcal {N}(varvec{0}, varvec{Sigma }), end{aligned}$$
(3)
where (g(cdot )) is a link function, (varvec{psi }) are additional parameters for the data model, and (varvec{Sigma }) is the covariance matrix of the spatial random effect. In the SGLMM, prior information facilitates the estimation of (varvec{eta },) which would not be estimable otherwise due to its shared column space with (varvec{beta })30. This is analogous to applying a ridge penalty to (varvec{eta },) which stabilizes the likelihood. Another method for fitting the confounded SGLMM is to specify a restricted version:$$begin{aligned} varvec{y}&sim [varvec{y}|varvec{mu }, varvec{psi }], end{aligned}$$
(4)
$$begin{aligned} g(varvec{mu })&= varvec{X}varvec{delta } + (varvec{I}-varvec{P}_{varvec{X}})varvec{eta }, end{aligned}$$
(5)
$$begin{aligned} varvec{eta }&sim mathcal {N}(varvec{0}, varvec{Sigma }), end{aligned}$$
(6)
where (varvec{P}_{varvec{X}}=varvec{X}(varvec{X}varvec{X})^{-1}varvec{X}’) is the projection matrix onto the column space of (varvec{X}.) In the unrestricted SGLMM, the regression coefficients (varvec{beta }) and random effect (varvec{eta }) in (1) compete to explain variability in the latent mean (varvec{mu }) in the direction of (varvec{X})27. In the restricted model, however, all variability in the direction of (varvec{X}) is explained solely by the regression coefficients (varvec{delta }) in (4)31, and (varvec{eta }) explains residual variation that is orthogonal to (varvec{X}). We refer to (varvec{beta }) as the conditional effects because they depend on (varvec{eta }), and (varvec{delta }) as the unconditional effects.Restricted regression, as specified in (4), was proposed by Reich et al.28. Reich et al.28 described a disease-mapping example in which the inclusion of a spatial random effect rendered one covariate effect unimportant that was important in the non-spatial model. Spatial maps indicated an association between the covariate and response, making inference from the spatial model appear untenable. Reich et al.28 proposed restricted spatial regression as a method for recovering the posterior expectations of the non-spatial model and shrinking the posterior variances which tend to be inflated for the unrestricted SGLMM.Several modifications of restricted spatial regression have been proposed30,32,33,34,35. All restricted spatial regression methods seek to provide posterior means (text {E}left( delta _j|varvec{y}right)) and marginal posterior variances (text {Var}left( delta _j|varvec{y}right)), (j=1,…,p) that satisfy the following two conditions36:

1.

(text {E}left( varvec{delta }|varvec{y}right) = text {E}left( varvec{beta }_{text {NS}}|varvec{y}right)) and,

2.

(text {Var}left( beta _{text {NS,}j}|varvec{y}right) le text {Var}left( delta _{j}|varvec{y}right) le text {Var}left( beta _{text {Spatial,}j}|varvec{y}right)) for (j=1,…,p),

where (varvec{beta }_{NS}) and (varvec{beta }_{Spatial}) are the regression coefficients corresponding to the non-spatial and unrestricted spatial models, respectively.The inferential impacts of spatial confounding on the regression coefficients has been debated. Hodges and Reich29 outlined five viewpoints on spatial confounding and restricted regression in the literature and refuted the two following views:

1.

Adding the random effect (varvec{eta }) corrects for bias in (varvec{beta }) resulting from missing covariates.

2.

Estimates of (varvec{beta }) in a SGLMM are shrunk by the random effect and hence conservative.

The random effect (varvec{eta }) can increase or decrease the magnitude of (varvec{beta }), and the change may be galvanized by mechanisms not related to missing covariates. Therefore, we cannot assume the regression coefficients in the SGLMM will exceed those of the restricted model, nor should we regard the estimates in either model as biased due to misspecification. Confounding in the SGLMM causes (text {Var}left( beta _j|varvec{y}right) ge text {Var}left( delta _j|varvec{y}right)), (j=1,…,p), because of the shared column space of the fixed and random effects. Thus, we refer to the conditional coefficients as conservative with regard to their credible intervals, not their posterior expectations.Reich et al.28 argued that restricted spatial regression should always be applied because the spatial random effect is generally added to improve predictions and/or correct the fixed effect variance estimate. While it may be inappropriate to orthogonalize a set of fixed effects in an ordinary linear model, orthogonalizing the fixed and random effect in a spatial model is permissible because the random effect is generally not of inferential interest. Paciorek37 provided the alternative perspective that, if confounding exists, it is inappropriate to attribute all contested variability in (varvec{y}) to the fixed effects. Hanks et al.31 discussed factors for deciding between the unrestricted and restricted SGLMM on a continuous spatial support. The restricted SGLMM leads to improved computational stability, but the unconditional effects are less conservative under model misspecification and more prone to type-S errors: The Bayesian analogue of Type I error. Fitting the unrestricted SGLMM when the fixed and random effects are truly orthogonal does not introduce bias, but it will increase the fixed effect variance. Given these considerations, Hanks et al.31 suggested a hybrid approach where the conditional effects, (varvec{beta }), are extracted from the restricted SGLMM. This is possible because the restricted SGLMM is a reparameterization of the unrestricted SGLMM. This hybrid approach leads to improved computational stability but yields the more conservative parameter estimates. We describe how to implement this hybrid approach for joint species distribution models in the “Community confounding” section.Restricted regression has also been applied in time series applications. Dominici et al.38 debiased estimates of fixed effects confounded by time using restricted smoothing splines. Without the temporal random effect, Dominici et al.38 asserted all temporal variation in the response would be wrongly attributed to temporally correlated fixed effects. Houseman et al.39 used restricted regression to ensure identifiability of a nonparametric temporal effect and highlighted certain covariate effects that were more evident in the restricted model (i.e., the unconditional effects’ magnitude was greater). Furthermore, restricted regression is implicit in restricted maximum likelihood estimation (REML). REML is often employed for debiasing the estimate of the variance of (varvec{y}) in linear regression and fitting linear mixed models that are not estimable in their unrestricted format40. Because REML is generally applied in the context of variance and covariance estimation, considerations regarding the effects of REML on inference for the fixed effects are lacking in the literature.In ecological science, JSDMs often include an unstructured random effect like (varvec{eta }) in (1) to account for interspecies dependence, and hence can also experience community confounding between (varvec{X}) and (varvec{eta }) analogous to spatial confounding. Unlike a spatial or temporal random effect, we consider random species effects to be inferentially important, rather than a tool solely for improving predictions or catch-all for missing covariates. An orthogonalization approach in a JSDM attributes contested variation between the fixed effects (environmental information) and random effect (community information) to the fixed effect.We describe how to orthogonalize the fixed and random species effects in a suite of JSDMs and present a method for detecting community confounding. In the simulation study, we test the efficacy of our method for detecting confounding, show that community confounding can lead to computational difficulties similar to those caused by spatial confounding31, and highlight that, for some models, restricted regression can improve model fitting. We also investigate the inferential implications of community confouding and restricted regression in JSDMs by comparing outputs from the SDM, unrestricted JSDM, and restricted JSDM of the Royle-Nichols and probit occupancy models fit to mammalian camera trap data. Lastly, we discuss other inferential and computational methods for confounded models and consider their appropriateness for joint species distribution modeling. More

This section is devoted to show results and to highlight eventual effects of the interplay between the nonlinearity characterizing the system dynamics and the presence of noisy fluctuations for the irradiance variable.Analysis of experimental dataThe need of taking into account noisy fluctuations of such an environmental variable is well demonstrated in Fig. 1. In the first panel (a) the experimental time behaviour of the irradiance is shown. This noisy curve is based on the experimental data (purple points) of the Boussole buoy located in the Gulf of Lion, collected over a period of nine years, precisely from 2004 to 2013. The time series of the experimental data presents quite a few gaps in time due to the malfunction of the buoy. This aspect has been remedied by merging the experimental data with those of the OASIM model validated for the Boussole site61 (yellow points). The latter is a multispectral atmospheric radiative transfer model that is in turn forced by experimental-model data based on ECMWF ERAINTERIM reanalyses which provide, for example, cloud cover data. The radiative model is partly stochastic since it considers the effects stemming from the presence of clouds, averaged along a single day (this explains why the yellow points are slightly less scattered). We see that the OASIM model accurately reproduces the profile which emerges from the experimental data. Further, we stress that the experimental data are only used in this initial analysis. In the biogeochemical simulations the irradiance signal is fully reconstructed starting from a realistic seasonal cycle combined with a range of different random fluctuations, and the information from OASIM is not used. In the second panel (b) the daily (black points) as well as the three-month (red points) running mean of the experimental series are plotted. Figure 1c shows the irradiance noisy fluctuations (INF) which have been obtained by subtracting the three-month running mean curve (3MRM, red curve in Fig. 1b) from the daily running mean one (DRM, black curve in Fig. 1b) and normalizing with respect to the mean of the 3MRM ((overline{3MRM})), namely (INF = (DRM – 3MRM) / overline{3MRM}). We see that a seasonal overall trend with higher oscillations during the winter time can be seen, implying that the characteristics of the noise may change over the year. Moreover, a slight imbalance between positive and negative values of the noisy fluctuations (that is, different values of the maximum fluctuation intensity) is present. The physical reason for the occurrence of such an aspect can be ascribed to the fact that the maximum value of solar irradiance corresponds to that measured during a sunny day. Conversely, the minimum level tends to zero corresponding to a dense darkness. While the former is close to the mean value of the solar irradiance (most of all in summer), the latter is much further away and then a natural asymmetry arises in the random fluctuations. However, it should be noted that, apart from the intense spikes, the asymmetry is not so pronounced, as proved by the mean value (red line in Fig. 1c) which is practically zero, namely (0.4%) of the (overline{3MRM}). Therefore, basing on this last observation, to model the noise affecting the irradiance dynamics, as a first approximation we consider a symmetric Gaussian autocorrelated noise as described in the next subsection.On the basis of such experimental results, we postulate the hypothesis that random fluctuations of light cannot be neglected, most of all in the study of ecological systems where light profoundly determines the system dynamics, governing fundamental processes at the basis of of the food web.Figure 1(a) Experimental data (purple points) of the stochastic solar irradiance collected by the Boussole buoy in a time-window of 9 years (2004-2013); the yellow points are the data generated by the OASIM model used to fill the gaps present in the experimental time-series due to malfunctioning of the buoy. (b) Daily (black points) and three-month (red points) running mean of the light curve in panel (a). (c) Irradiance noisy fluctuations (INF), obtained by subtracting the three-month running mean curve (3MRM) from the daily running mean one (DRM) and normalizing with respect to the mean value of 3MRM ((overline{3MRM})), namely (INF = (DRM – 3MRM) / overline{3MRM}); the red line represents the mean value of such fluctuations. Data already presented and validated in61.Full size imageSolar irradianceThe solar irradiance forcing is derived considering a deterministic seasonal oscillation combined with an Ornstein-Uhlenbeck process. The coefficient of variation (CV) of simulated light forcing, Fig. 2, (CV=sigma / mu) ((mu) and (sigma) being mean value and standard deviation calculated over both time and numerical realizations), is shown for 231 (D-tau) pairs. D and (tau) represent the intensity of a Gaussian noise source and the auto-correlation time of the fluctuations, respectively (see Eqs. (2) and (3)).Each pixel represents the mean value on time of CV calculated with respect to 1000 different stochastic realizations. Figure 2Coefficient of variation ((CV=sigma / mu)) of irradiance resulting from numerical integration of model equations for 231 (D-tau) different scenarios.Full size imageIt is easy to see the agreement between the results obtained from the numerical integration and the theoretical ones derivable from Eq. (5) by putting (text {var}{F_L(0)}=0) and (t gg 1), getting (sigma ^2_L=D / 2tau). In Fig. 2, indeed, the maximum values of (sigma) lie in the upper left part of the plot corresponding to small (high) values of (tau) (D). As it is clear the values of D have been chosen in order to obtain a relative standard deviation ranging from (5%mu) to (60%mu). We underline that, in this case, it is possible to interchangeably consider (sigma) and CV since the dependence of CV on D and (tau) does not differ from that of (sigma) (meaning that the dependence of (sigma) is not altered by dividing by (mu)) (results not shown).Effects on population dynamicsIn this section the noise-induced effects on the population dynamics are examined. The nine planktonic populations present a different qualitative behaviour of the CV, compared to that of the irradiance. In this case, the CV is characterized by a strong non-monotonic dependence on the parameter (tau). This aspect can be appreciated in Fig. 3 where different curves of CV versus the time correlation parameter are shown for different fixed values of D.Figure 3Coefficient of variation ((CV=sigma / mu)) of the nine planktonic populations resulting from numerical integration of model equations plotted versus the considered values of (tau); the different curves are related to different values of the noise intensity D.Full size imageThe existence of a maximum value for CV can be appreciated for each species. Although the qualitative behaviour is the same for all strains, particular attention has to be payed on diatoms and nanoflagellates. All the other species, indeed, present a percent variation of standard deviation between (2%) and (15%). In the case of nanoflagellates, instead, the D-dependent range is (20-90%), while diatoms reach values over the (100%) for the highest values of D. Therefore, these two species, in particular, and the whole system, in general, are extremely sensitive to the auto-correlation time which characterizes the noise.We note that the different curves related to the different selected values of D approach the horizontal axis, tending asymptotically to vanish as (tau) increases. Such a behaviour can be explained by the fact that high values of (tau) give rise to a more correlated dynamics, so that (tau rightarrow infty) implies fully correlated time-behaviours corresponding to the deterministic case. In this instance, then, all the different realizations give the same results, making the standard deviation vanish. The same happens, independently of the value of (tau), for low values of noise intensity for which the corresponding curves approach the same almost vanishing value (see orange, gray and yellow lines). Differently from the previous case, when (tau rightarrow 0) the noise tends to a delta-correlated noise, that is a white noise; for (tau ne 0), instead, the noise spectrum is not flat, being characterized by a Cauchy-Lorentz distribution. The strong nonmonotonicity of CV with respect to (tau), emerging when there are relatively high values of CV, implies a greater variability of the system biomass. Lower values of CV indicate that the system dynamics is less influenced by the presence of noise where very little or no differences with respect to the deterministic case are present. Conversely, high values of CV clearly demonstrate the remarkable signature of the presence of an impacting noise source. It is interesting to note that the noise influence on the ecosystem strongly depends on both (tau) and D, that is, just an intense noise is not enough to generate a greater response of the ecosystem. In particular, experimental data are characterized by a CV approximately equal to 0.361, which corresponds to values of D and (tau) lying on the diagonal strip in Fig. 2 ranging from ((tau ,D)=(0.5,10^4)) to ((tau ,D)=(365,10^7)). Finally we note the presence of a noise suppression effect. High values of D, indeed, can generate slight effects when the correlation time (tau) does not take on suitable values.The results shown here are an extension of the previous work by Benincà et al.56. There, the authors analyse a simpler, less realistic model of two interacting populations, whose dynamics is affected by a randomly fluctuating temperature. In that case, moreover, the deterministic oscillations of the temperature are suppressed, and the system exhibits intrinsic Lotka-Volterra oscillations whose frequency match with the characteristic one(s) of the noise. On the contrary, here, the observed maximum response (see Fig. 3) cannot be interpreted as a synchronization effect, since our model does not present intrinsic Lotka-Volterra-like oscillations and the periodic population variability is only due to the deterministic forcing(s).The nonmonotonic behaviour of the CV can be then interpreted as the signature of the intimate interplay between the ecological system and the noise. This interplay, indeed, has a pivotal role in both determining the dynamics of the populations and defining the characteristics of the ecosystem.In Fig. 3 it can be observed that the value of (tau) for which CV is maximum strongly depends on the noise intensity D. In particular, it is possible to note that the peaks in Fig. 3 move towards higher values of (tau) as the noise intensity increases. Thus, Fig. 3 demonstrates that the maximum-response effect to the random fluctuations is sensitive to the noise intensity D.However, it is important to underline that the response of the system to the noisy signal does not depend on the yearly oscillations induced by the deterministic forcings. Indeed, by considering constant the deterministic part of all external forcings (temperature, irradiance, wind and salinity), the non monotonic behaviour of CV with respect to both (tau) and D is still present, provided that the populations are not extinct (plot not shown). In this scenario indeed, besides dinoflagellates, diatoms and nanoflagellates are practically extinct as well, exhibiting thus a constant vanishing variance. All the other strains, instead, present qualitatively the same nonmonotonicity with only slight differences (shift of the peaks and different mean values of the CV curves), probably due to the extinction of diatoms and nanoflagellates which causes relevant differences in the system dynamics. More specifically, the system’s response seems to depend on both the noise intensity and the correlation time (see Fig. 3).In this scenario (absence of seasonal driving) we have studied the dependence on both parameters D and (tau) of the probability density functions (PDFs) of the non-vanishing populations. In Fig. 4, the PDFs of bacteria (B1), picophytoplankton (P3), microzooplankton (Z5) and etherotrophic nanoflagellates (Z6) are plotted for (tau =0.5) and eight different values of the parameter D.Figure 4Dependence of the probability density functions of non-vanishing populations on the parameter D for (tau =0.5). The curves are normalized within the interval taken into account. For this reason the relative peaks of the curves in the bottom panels have different values compared to those of the top panels. However, the figure aims at showing the existence of the value of the noise intensity for which the system is more sensitive as well as the generation of a stationary out-of-equilibrium state induced by the noise.Full size imageWe see that the mean value and the variance of these populations are strongly affected by the presence of random fluctuations in the irradiance. Specifically, as the noise intensity increases the mean values of picophytoplankton and bacteria concentrations exhibit a shift. In particular, the results indicate that picophytoplankton is disavantaged by the presence of a noisy component in the irradiance, which indeed tends to inhibit its ability to absorbe the solar light, slowing down its growth. As a consequence, since phytoplankton and bacteria compete for the same resources, as the former declines the latter are favoured, with a compensation mechanism which allows their predators (zooplankton populations) to be almost not affected by the noisy behaviour of the irradiance. Further, we note that for intermediate values of the noise intensity ((D = 10^4 – 10^5)) a maximum of the variance occurs (the PDFs are clearly spread on a wider range of values). Such an effect indicates that the noisy behaviour of irradiance strongly influences the whole ecosystem dynamics. Moreover, the nonmonotonic behaviour of the variance (its PDFs become larger and then tighter again as the noise intensity increases) indicates that the noise pushes the ecosystem away from equilibrium, driving it towards a non-equilibrium steady state. Finally, we note that the nonmonotonic behaviour of CV as a function of the noise intensity remains also in the presence of seasonal driving.Figure 5Coefficient of variation ((CV=sigma / mu)) of nine planktonic populations resulting from numerical integration of model equations plotted versus the considered values of D; different curves correspond to different values of the correlation time (tau).Full size imageFigure 5 shows indeed the nonmonotonic response of the ecosystem to the change of D when the deterministic seasonal cycling of the four environmental parameters (temperature, irradiance, wind and salinity) is present. It is easy to observe that also in this instance the major noise-induced effect appears in nanoflagellates and diatoms with a percent standard deviation of 50(%) and 100(%), respectively. The coalescence of different curves (related to different values of (tau)), as D decreases, is due to the fact that for (D rightarrow 0) the impact of the noise is negligible and the evolution of the system practically resembles the deterministic one. On the contrary, for higher values of D remarkable differences arise and clear peaks of CV appear in the considered range of variation.These plots show that, for a fixed value of (tau), there exists a value of the noise intensity for which the planktonic concentrations are maximally spread around their mean values (corresponding to the maximum value of CV and then of the variance). Moreover, such a nonmonotonic behaviour suggests the presence of a resonance, which can be interpreted as the effect of the interplay between the nonlinearity of the system and the environmental random fluctuations.Also in this case, the interplay between the two parameters D and (tau) in determining and characterizing the dynamics of the ecosystem transparently emerges. The value of D corresponding to the maximum value of CV, indeed, basically depends on the specific value of (tau).Finally, we point out that the different dynamic scenarios identified by the D-(tau) couples can be experienced by the system during the year, since the two parameters may seasonally vary depending on the different weather conditions. In other words, a seasonally varying noise (see Fig. 1c) may cause the nine populations explore different regions of the D-(tau) space during the year. Therefore, the results reported in this paper can highlight the detectable yearly variability of a marine ecosystem which does not stem from the deterministic seasonal variation of environmental parameters.Effects on the organic carbonIn this subsection the effects of the irradiance noise on the biogechemistry are analysed. In Fig. 6 the dependence on (tau) of both the CV [panel (a)] and the mean value concentration [panel (b)] of detritus, labile dissolved organic carbon (L-DOC), semi-labile dissolved organic carbon (SL-DOC) and gross primary production (GPP) are shown. All these biogeochemical properties are correlated with carbon cycling. Gross primary production is related to the amount of carbon entering in the ecosystem, and is related to the maximum energy available in the ecosystem progressively dissipated in the trophic web. Gross primary production is directly affected by light fluctuation and its CV shape is very similar to that of the irradiance, Fig. 2. We selected also detritus and DOC because they are important indicators for the carbon cycling dynamics and are related to the cycling of chemicals like heavy metals62. The different curves, related to different values of D, approach the same (vanishing) value for large (tau). As previously discussed for the CV [Fig. 6(a)] of biomass concentrations, this circumstance is due to the fact that, in this case, the system dynamics tends to the deterministic case, characterized by a unique possible realization implying a vanishing standard deviation. For high correlation times thus the system is insensitive to the noise intensity. On the contrary, for small values of (tau), different values of D lead to significant differences of the variance. In particular, detritus, L-DOC and SL-DOC exhibit a clear non-monotonic behaviour whose maximum value depends on the combined values of D-(tau). Only the GPP presents a decreasing monotonic behaviour.The dependence of the mean value concentration on (tau), instead, is qualitatively the same for all the four parameters. Also in this case we can note a diversification with respect to D occurring at small (tau) and a (deterministic) constant value arising for low (high) values of D ((tau)).These results manifest that not only the population dynamics, but also all the biogeochemical processes are profoundly affected by the presence of stochastic environmental variables. The values and the behaviour of the examined quantities are indeed determined by the intimate interplay between the intensity and the time correlation of the noise fluctuations.Figure 6(a) Coefficient of variation ((CV=sigma / mu)) and (b) mean value concentration ((mu)) of detritus, labile dissolved organic carbon (L-DOC), semi-labile dissolved organic carbon (SL-DOC) and gross primary production (GPP) resulting from numerical integration of model equations plotted versus the considered values of (tau); the different curves are related to different values of the correlation time D.Full size image More

Armbrust, E. V. The life of diatoms in the world’s oceans. Nature 459, 185–192 (2009).CAS 
Article 

Google Scholar 
Mann, D. G. The species concept in diatoms. Phycologia 38, 437–495 (1999).Article 

Google Scholar 
Smetacek, V. Diatoms and the ocean carbon cycle. Protist 150, 25–32 (1999).CAS 
Article 

Google Scholar 
Rynearson, T. A. et al. Major contribution of diatom resting spores to vertical flux in the sub-polar North Atlantic. Deep. Res. Part I Oceanogr. Res. Pap. 82, 60–71 (2013).CAS 
Article 

Google Scholar 
Allen, J. T. et al. Diatom carbon export enhanced by silicate upwelling in the northeast Atlantic. Nature 437, 728–732 (2005).CAS 
Article 

Google Scholar 
Boyd, P. W., Strzepek, R., Fu, F. & Hutchins, D. A. Environmental control of open-ocean phytoplankton groups: Now and in the future. Limnol. Oceanogr. 55, 1353–1376 (2010).CAS 
Article 

Google Scholar 
Hátún, H., Somavilla, R., Rey, F., Johnson, C. & Mathis, M. The subpolar gyre regulates silicate concentrations in the North Atlantic. Sci. Rep. 1–9 https://doi.org/10.1038/s41598-017-14837-4 (2017).Bopp, L. Response of diatoms distribution to global warming and potential implications: a global model study. Geophys. Res. Lett. 32, 2–5 (2005).Article 
CAS 

Google Scholar 
Warner, A. J. & Hays, G. C. Sampling by the Continuous Plankton Recorder survey. Prog. Oceanogr. 6611, 237–256 (1994).Article 

Google Scholar 
Edwards, M., Beaugrand, G., Reid, P. C., Rowden, A. A. & Jones, M. B. Ocean climate anomalies and the ecology of the North Sea. Mar. Ecol. Prog. Ser. 239, 1–10 (2002).Article 

Google Scholar 
Allen, S. et al. Interannual stability of phytoplankton community composition in the North-East atlantic. Mar. Ecol. Prog. Ser. 655, 43–57 (2020).Article 

Google Scholar 
Edwards, M. & Richardson, A. J. Impact of climate change on marine pelagic phenology and trophic mismatch. Nature 430, 881–884 (2004).CAS 
Article 

Google Scholar 
Wihsgott, J. U. et al. Observations of vertical mixing in autumn and its effect on the autumn phytoplankton bloom. Prog. Oceanogr. 177, 1157–1165 (2019).Article 

Google Scholar 
Kamykowski, D. & Zentara, S. J. Predicting plant nutrient concentrations from temperature and sigma-T in the upper kilometer of the world ocean. Deep. Res. Part A-Oceanogr. Res. Pap. 33, 89–105 (1986).CAS 
Article 

Google Scholar 
Kamykowski, D. & Zentara, S. J. Changes in world ocean nitrate availability through the 20th century. Deep. Res. Part I-Oceanogr. Res. Pap. 52, 1719–1744 (2005).Article 

Google Scholar 
Behrenfeld, M. J. et al. Climate-driven trends in contemporary ocean productivity. Nature 444, 752–755 (2006).CAS 
Article 

Google Scholar 
López-Urrutia, A., San Martin, E., Harris, R. P. & Irigoien, X. Scaling the metabolic balance of the oceans. Proc. Natl. Acad. Sci. USA 103, 8739–8744 (2006).Article 
CAS 

Google Scholar 
Richardson, A. J. & Schoeman, D. S. Climate impact on plankton ecosystems in the Northeast Atlantic. Science 305, 1609–1612 (2004).CAS 
Article 

Google Scholar 
Edwards, M., Johns, D., Leterme, S., Svendsen, E. & Richardson, A. Regional climate change and harmful algal blooms in the northeast Atlantic. Limnol. Oceanogr. 51, 820–829 (2006).Article 

Google Scholar 
Hinder, S. L. et al. Changes in marine dinoflagellate and diatom abundance under climate change. Nat. Clim. Chang. 2, 271–275 (2012).Article 

Google Scholar 
Batten, S. et al. CPR sampling: the technical background, materials and methods, consistency and comparability. Prog. Oceanogr. 58, 193–215 (2003).Article 

Google Scholar 
Reid, P. C. et al. The Continuous Plankton Recorder: concepts and history, from plankton indicator to undulating recorders. Prog. Oceanogr. 58, 117–173 (2003).Article 

Google Scholar 
Edwards, M., Beaugrand, G., Hays, G. C., Koslow, J. A. & Richardson, A. J. Multi-decadal oceanic ecological datasets and their application in marine policy and management. Trends Ecol. Evol. 25, 602–610 (2010).Article 

Google Scholar 
Edwards, M. et al. North Atlantic warming over six decades drives decreases in krill abundance with no associated range shift. Commun. Biol. 4, 1–10 (2021).Article 

Google Scholar 
Richardson, A. J. et al. Using continuous plankton recorder data. Prog. Oceanogr. 68, 27–74 (2006).Article 

Google Scholar 
Hélaouët, P., Beaugrand, G. & Reygondeau, G. Reliability of spatial and temporal patterns of C. finmarchicus inferred from the CPR survey. J. Mar. Syst. 153, 18–24 (2016).Article 

Google Scholar 
Owens, N. J. P. et al. All plankton sampling systems underestimate abundance: response to “Continuous plankton recorder underestimates zooplankton abundance” by J.W. Dippner and M. Krause. J. Mar. Syst. 128, 240–242 (2013).Article 

Google Scholar 
Jonas, T. D., Walne, A., Beaugrand, G., Gregory, L. & Hays, G. C. The volume of water filtered by a Continuous Plankton Recorder sample: the effect of ship speed. J. Plankton Res. 26, 1499–1506 (2004).Article 

Google Scholar 
O’Reilly, C. H., Zanna, L. & Woollings, T. Assessing external and internal sources of Atlantic multidecadal variability using models, proxy data, and early instrumental indices. J. Clim 32, 7727–7745 (2019).Article 

Google Scholar 
Qin, M., Dai, A. & Hua, W. Quantifying contributions of internal variability and external forcing to atlantic multidecadal variability since 1870. Geophys. Res. Lett. 47, 1–11 (2020).
Google Scholar 
Mann, M. E., Steinman, B. A. & Miller, S. K. Absence of internal multidecadal and interdecadal oscillations in climate model simulations. Nat. Commun. 11, 1–9 (2020).Article 
CAS 

Google Scholar 
Enfield, D. B., Mestas-Nuñez, A. M. & Trimble, P. J. The Atlantic Multidecadal Oscillation and its relation to rainfall and river flows in the continental U.S. Geophys. Res. Lett. 28, 2077–2080 (2001).Article 

Google Scholar 
Gray, S. T., Graumlich, L. J., Betancourt, J. L. & Pederson, G. T. A tree-ring based reconstruction of the Atlantic Multidecadal Oscillation since 1567 A.D. Geophys. Res. Lett. 31, 2–5 (2004).Article 

Google Scholar 
Edwards, M., Beaugrand, G., Helaouët, P., Alheit, J. & Coombs, S. Marine ecosystem response to the Atlantic Multidecadal Oscillation. PLoS One 8, e57212 (2013).CAS 
Article 

Google Scholar 
Jones, P. D., New, M., Parker, D. E. & Martin, S. & Rigor, I. G. Surface air temperature and its changes over the past 150 years. Rev. Geophys. 37, 173–199 (1999).Article 

Google Scholar 
Beaugrand, G. et al. Reorganization of North Atlantic marine copepod biodiversity and climate. Science296, 1692–1694 (2002).CAS 
Article 

Google Scholar 
Alvain, S., Moulin, C., Dandonneau, Y. & Bréon, F. M. Remote sensing of phytoplankton groups in case 1 waters from global SeaWiFS imagery. Deep Sea Res. Part I Oceanogr. Res. Pap. 52, 1989–2004 (2005).Article 

Google Scholar 
Kalnay, E. et al. The NCEP/NCAR 40-year reanalysis project. Bull. Am. Meteorol. Soc. 77, 437–471 (1996).Article 

Google Scholar 
Huang, B. et al. Extended reconstructed sea surface temperature, version 5 (ERSSTv5): upgrades, validations, and intercomparisons. J. Clim. 30, 8179–8205 (2017).Article 

Google Scholar 
Lam, N. S. N. Spatial interpolation methods: a review. Am. Cartogr. 10, 129–150 (1983).Article 

Google Scholar 
Beaugrand, G., McQuatters-Gollop, A., Edwards, M. & Goberville, E. Long-term responses of North Atlantic calcifying plankton to climate change. Nat. Clim. Chang. 3, 263–267 (2012).Article 
CAS 

Google Scholar 
Beaugrand, G. et al. Prediction of unprecedented biological shifts in the global ocean. Nat. Clim. Chang. 9, 237–243 (2019).Article 

Google Scholar  More

Chapron, G. et al. Recovery of large carnivores in Europe’s modern human-dominated landscapes. Science 346, 1517–1519 (2014).ADS 
CAS 
PubMed 
Article 

Google Scholar 
Lute, M. L., Carter, N. H., López-Bao, J. V. & Linnell, J. D. C. Conservation professionals’ views on governing for coexistence with large carnivores. Biol. Cons. 248, 108668 (2020).Article 

Google Scholar 
Gantchoff, M. G. & Belant, J. L. Regional connectivity for recolonizing American black bears (Ursus americanus) in southcentral USA. Biol. Cons. 214, 66–75 (2017).Article 

Google Scholar 
Ripple, W. J. et al. Status and ecological effects of the world’s largest carnivores. Science 343, 25 (2014).Article 
CAS 

Google Scholar 
Kays, R. et al. Does hunting or hiking affect wildlife communities in protected areas?. J. Appl. Ecol. 54, 242–252 (2017).Article 

Google Scholar 
Schipper, J. et al. The status of the world’s land and marine mammals: diversity, threat, and knowledge. Science 322, 225–230 (2008).ADS 
CAS 
PubMed 
Article 

Google Scholar 
Smith, J. A., Wang, Y. & Wilmers, C. C. Top carnivores increase their kill rates on prey as a response to human-induced fear. Proc. R. Soc. B Biol. Sci. 282, 20142711 (2015).Article 

Google Scholar 
Stillfried, M., Belant, J. L., Svoboda, N. J., Beyer, D. E. & Kramer-Schadt, S. When top predators become prey: Black bears alter movement behaviour in response to hunting pressure. Behav. Proc. 120, 30–39 (2015).Article 

Google Scholar 
Støen, O.-G. et al. Physiological evidence for a human-induced landscape of fear in brown bears (Ursus arctos). Physiol. Behav. 152, 244–248 (2015).PubMed 
Article 
CAS 

Google Scholar 
Evans, M. J., Rittenhouse, T. A. G., Hawley, J. E. & Rego, P. W. Black bear recolonization patterns in a human-dominated landscape vary based on housing: New insights from spatially explicit density models. Landsc. Urban Plan. 162, 13–24 (2017).Article 

Google Scholar 
LaRue, M. A. et al. Cougars are recolonizing the midwest: Analysis of cougar confirmations during 1990–2008. J. Wildl. Manag. 76, 1364–1369 (2012).Article 

Google Scholar 
Cove, M. V., Fergus, C., Lacher, I., Akre, T. & McShea, W. J. Projecting mammal distributions in response to future alternative landscapes in a rapidly transitioning region. Remote Sens. 11, 2482 (2019).ADS 
Article 

Google Scholar 
Frid, A. & Dill, L. Human-caused disturbance stimuli as a form of predation risk. Conserv. Ecol. 6, 25 (2002).
Google Scholar 
Clinchy, M. et al. Fear of the human “super predator” far exceeds the fear of large carnivores in a model mesocarnivore. Behav. Ecol. 27, 1826–1832 (2016).
Google Scholar 
Suraci, J. P., Clinchy, M., Zanette, L. Y. & Wilmers, C. C. Fear of humans as apex predators has landscape-scale impacts from mountain lions to mice. Ecol. Lett. 22, 1578–1586 (2019).PubMed 
Article 

Google Scholar 
Gaynor, K. M., Hojnowski, C. E., Carter, N. H. & Brashares, J. S. The influence of human disturbance on wildlife nocturnality. Science 360, 1232–1235 (2018).ADS 
CAS 
PubMed 
Article 

Google Scholar 
Smith, J. A., Thomas, A. C., Levi, T., Wang, Y. & Wilmers, C. C. Human activity reduces niche partitioning among three widespread mesocarnivores. Oikos 127, 890–901 (2018).Article 

Google Scholar 
Tucker, M. A. et al. Moving in the Anthropocene: Global reductions in terrestrial mammalian movements. Science 359, 466–469 (2018).ADS 
CAS 
PubMed 
Article 

Google Scholar 
Carter, N. H., Brown, D. G., Etter, D. R. & Visser, L. G. American black bear habitat selection in northern Lower Peninsula, Michigan, USA, using discrete-choice modeling. Ursus 21, 57–71 (2010).Article 

Google Scholar 
Naidoo, R. & Burton, A. C. Relative effects of recreational activities on a temperate terrestrial wildlife assemblage. Conserv. Sci. Pract. 2, e271 (2020).
Google Scholar 
Geffroy, B., Samia, D. S. M., Bessa, E. & Blumstein, D. T. How nature-based tourism might increase prey vulnerability to predators. Trends Ecol. Evol. 30, 755–765 (2015).PubMed 
Article 

Google Scholar 
Geffroy, B. et al. Evolutionary dynamics in the Anthropocene: Life history and intensity of human contact shape antipredator responses. PLoS Biol. 18, e3000818 (2020).CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Beeco, J. A., Hallo, J. C. & Brownlee, M. T. J. GPS visitor tracking and recreation suitability mapping: tools for understanding and managing visitor use. Landsc. Urban Plan. 127, 136–145 (2014).Article 

Google Scholar 
Thorsen, N. H. et al. Smartphone app reveals that lynx avoid human recreationists on local scale, but not home range scale. Sci. Rep. 12, 1–13 (2022).Article 
CAS 

Google Scholar 
Evans, M. J., Hawley, J. E., Rego, P. W. & Rittenhouse, T. A. G. Hourly movement decisions indicate how a large carnivore inhabits developed landscapes. Oecologia 190, 11–23 (2019).ADS 
PubMed 
Article 

Google Scholar 
Carlos, A. W. D., Bright, A. D., Teel, T. L. & Vaske, J. J. Human-black bear conflict in urban areas: an integrated approach to management response. Hum. Dimens. Wildl. 14, 174–184 (2009).Article 

Google Scholar 
Johnson, H. E. et al. Human development and climate affect hibernation in a large carnivore with implications for human–carnivore conflicts. J. Appl. Ecol. 55, 663–672 (2018).Article 

Google Scholar 
Gould, N. P., Powell, R., Olfenbuttel, C. & DePerno, C. S. Growth and reproduction by young urban and rural black bears. J. Mammal. 102, 1165–1173 (2021).Article 

Google Scholar 
Ditmer, M. A., Noyce, K. V., Fieberg, J. R. & Garshelis, D. L. Delineating the ecological and geographic edge of an opportunist: The American black bear exploiting an agricultural landscape. Ecol. Model. 387, 205–219 (2018).Article 

Google Scholar 
McFadden-Hiller, J. E. Jr. & Belant, J. L. Spatial distribution of black bear incident reports in michigan. PLoS One 11, e0154474 (2016).PubMed 
PubMed Central 
Article 
CAS 

Google Scholar 
Ladle, A., Steenweg, R., Shepherd, B. & Boyce, M. S. The role of human outdoor recreation in shaping patterns of grizzly bear-black bear co-occurrence. PLoS One 13, e0191730 (2018).PubMed 
PubMed Central 
Article 

Google Scholar 
Wilbur, R. C., Lischka, S. A., Young, J. R. & Johnson, H. E. Experience, attitudes, and demographic factors influence the probability of reporting human–black bear interactions. Wildl. Soc. Bull. 42, 22–31 (2018).Article 

Google Scholar 
Lustig, E. J., Lyda, S. B., Leslie, D. M., Luttbeg, B. & Fairbanks, W. S. Resource selection by recolonizing American Black Bears. J. Wildl. Manage. 85, 531–542 (2021).Article 

Google Scholar 
Sun, C. C., Fuller, A. K., Hare, M. P. & Hurst, J. E. Evaluating population expansion of black bears using spatial capture-recapture. J. Wildl. Manage. 81, 814–823 (2017).Article 

Google Scholar 
Kautz, T. M. et al. Large carnivore response to human road use suggests a landscape of coexistence. Glob. Ecol. Conserv. 30, e01772 (2021).Article 

Google Scholar 
Michigan Department of Natural Resources (MIDNR) (2021).Blount, J. D., Chynoweth, M. W., Green, A. M. & Şekercioğlu, Ç. H. Review: COVID-19 highlights the importance of camera traps for wildlife conservation research and management. Biol. Cons. 256, 108984 (2021).Article 

Google Scholar 
Weather Atlas. https://www.weather-atlas.com/enEvans, J. S. Spatial Analysis and Modelling Utilities. Package ‘spatialEco’. https://cran.r-project.org/web/packages/spatialEco/spatialEco.pdf (2021).Díaz-Ruiz, F., Caro, J., Delibes-Mateos, M., Arroyo, B. & Ferreras, P. Drivers of red fox (Vulpes vulpes) daily activity: prey availability, human disturbance or habitat structure?. J. Zool. 298, 128–138 (2016).Article 

Google Scholar 
Moore, J. F. et al. Comparison of species richness and detection between line transects, ground camera traps, and arboreal camera traps. Anim. Conserv. 23, 561–572 (2020).Article 

Google Scholar 
Parsons, A. W. et al. Urbanization focuses carnivore activity in remaining natural habitats, increasing species interactions. J. Appl. Ecol. 56, 1894–1904 (2019).Article 

Google Scholar 
Allen, M. L., Sibarani, M. C., Utoyo, L. & Krofel, M. Terrestrial mammal community richness and temporal overlap between tigers and other carnivores in Bukit Barisan Selatan National Park, Sumatra. Anim. Biodiv. Conserv. 43(1), 97–107 (2020).Article 

Google Scholar 
Tian, C. et al. Temporal niche patterns of large mammals in Wanglang National Nature Reserve, China. Glob. Ecol. Conserv. 22, e01015 (2020).Article 

Google Scholar 
Meredith, M. & Ridout, M. Estimates of coefficient of overlapping for animal activity patterns. Package ‘overlap’. https://cran.r-project.org/web/packages/overlap/overlap.pdf (2020).RStudio Team. RStudio: Integrated Development for R. RStudio, PBC, Boston, MA. http://www.rstudio.com/ (2021).Ridout, M. S. & Linkie, M. Estimating overlap of daily activity patterns from camera trap data. JABES 14, 322–337 (2009).MathSciNet 
MATH 
Article 

Google Scholar 
Lashley, M. A. et al. Estimating wildlife activity curves: comparison of methods and sample size. Sci. Rep. 8, 1–11 (2018).CAS 
Article 

Google Scholar 
Rowcliffe, M. Animal Activity Statistics. Package ‘activity’. https://cran.r-project.org/web/packages/activity/activity.pdf (2021).MacKenzie, D. I., Nichols, J. D., Hines, J. E., Knutson, M. G. & Franklin, A. B. Estimating site occupancy, colonization, and local extinction when a species is detected imperfectly. Ecology 84, 2200–2207 (2003).Article 

Google Scholar 
Wei, T., & Simko, V. Visualization of a Correlation Matrix. Package ‘corrplot’. https://cran.r-project.org/web/packages/corrplot/corrplot.pdf (2017).Norton, D. C. et al. Female American black bears do not alter space use or movements to reduce infanticide risk. PLoS One 13, e0203651 (2018).PubMed 
PubMed Central 
Article 
CAS 

Google Scholar 
Ditmer, M. A. et al. Behavioral and physiological responses of American black bears to landscape features within an agricultural region. Ecosphere 6, 1–21 (2015).Article 

Google Scholar 
Clark, D. et al. Using machine learning methods to predict the movement trajectories of the Louisiana black bear. SMU Data Sci. Rev. 5, 25 (2021).
Google Scholar  More

Growth conditions and plant materialsTwo experiments were conducted concurrently (sites A and B) in the same screen house in 2019 between the months of May and July at the Landmark University Greenhouse and Hydroponic Technology Center, a section of the Teaching and Research Farm of the University in Omu-Aran, Kwara State Nigeria. Experiment at site B was conducted simultaneously as A so as to validate the results of experiment A. Landmark University lies within Latitude 8° 7′ 26.21388″ and 5° 5′ 0.1788″. Both experiments (A & B) involved tomato (Solanum lycopersicum L. variety cherry) and cucumber (Cucumis sativus L. variety marketer) crops. For each crop, seeds were sown into a separate seed tray filled with coco peat (Coco peat, SRIMATHI EXPORT, INDIA). Cocopeat is the mesocarp tissue or husk after the grinding of coconut fruit. It has a lightweight and high water and nutrient holding capacities, it has an acceptable pH, electrical conductivity, and other chemical attributes27. Rice husk is the by-product of rice after milling. The rice husk used was collected from the rice processing mill of Landmark University. Rice husk is a highly porous and light weighted material with a very high specific area28.Two sets of seed trays (one for organic and another for inorganic fertilizers) were used each for tomato and cucumber crops in the nursery. Both were raised in the nursery for two weeks before transplanting. Black grow bags (30 × 17 cm) filled with a coco peat/rice husk (1:4 ratio by volume) mixture with a weight of about 10 kg were arranged in a screen house. Both the nursery and establishment of crop proper take place in a screen house. The screen house has a galvanized iron as the frame, a UV covering on top, side net for screening insect pests the floor fairly covered with granite. Temperature and relative humidity within the screen house during the period of the experiment was monitored using a Thermograph and a Barograph, and they were at an average of 31 °C and 75%, respectively.The grow bags were randomly placed in the screen house for the unbiased application of amendments. For both tomato and cucumber crops, the treatment comprised of six (6) levels of liquid organic fertilizer (5, 15, 25, 35, 45, 55 mL), in-organic fertilizer, and a control (ordinary borehole water). Levels of organic fertilizers were selected based on the recommendation of 20 mL of liquid organic fertilizer by29. The eight (8) treatments both for tomato and cucumber were arranged in a Completely Randomized Design replicated three times. One healthy plant was maintained per grow bag and four grow bags represent a treatment and there were 32 plants per block each for tomato and cucumber. For both crops, the experiment lasted for 90 days.Organic and in-organic nutrient solutionsThe liquid organic fertilizer used was obtained from the biomass of Mexican sunflower (Tithonia diversifolia). Fresh biomass (mainly leaves and stems) of the plant was collected from the Teaching and Research Farms of Landmark University, Nigeria. After rinsing, they were cut with a sterile knife into pieces of ≤ 1 cm size. A sample was taken for initial physicochemical analyses by grinding in a sterile mortal, diluted with sterile water and analyzed. The biomass was then soaked in sterile water inside a clean container, and allowed to ferment spontaneously for a period of 14 days. During the fermentation, samples were taken every 4 days for microbial analyses of the major players during the fermentation. At the end of fermentation, the mixture was separated using a sieve of mesh size ≤ 2 mm. The liquid portion was then refrigerated prior to the planting regime while another sample was taken to ascertain the physicochemical and microbial qualities of the produced liquid fertilizer. The chemical analysis is presented in Table 4. For inorganic fertilizer, Water soluble fertilizers employed in hydroponics were used (Hydroponics fertilizer, Anmol chemicals, India); calcium nitrate 650 mg L−1, potassium nitrate 450 mg L−1, magnesium 400 mg L−1, chelate 20 mg L−1, mono-ammonium phosphate 400 mg L−1. The electrical conductivity (EC) of the solution was 1.9 dS m-1.Irrigation and fertigationThe tomato and cucumber plants were fertigated morning and evening daily for one hour on each occasion according to the treatments. Preparation of the nutrient solution was with borehole water and was supplied to plants by an online pressure drip irrigation system set at 2.0 L h-1 using an arrowhead on each tomato and cucumber plant. Different tanks (250 L) were installed according to the various treatments making a total of 8 tanks. The organic fertilizer was diluted according to the various treatments equivalent to 1.25, 3.75, 6.25, 8.75, 11.25, and 13.75 L per 250 L of water respectively for 5, 15, 25, 35, 45, and 55 mL treatments. The nutrient solutions were refilled when the consumption is less than 20% of the initial volume (250 L) in the tank. One day per week, crops were irrigated with ordinary water to wash out pipes and prevent deposits of salts. The same concentration of nutrient was used from transplanting to the termination of the study for both tomato and cucumber crops, however, at the flowering of the crops, the volume of fertigation was increased to 3.0 L h-1 to be able to cope with the size of the plants.Trellising, pest and diseases controlFor both tomato and cucumber crops, plant vines were supported by twisting them around a wire that is- attached to the roof of the screen house and 2 m from the ground. Lateral outgrowths were cut off every week to ensure a sturdy single stem. Pests and diseases were scouted every day. Whiteflies, aphids, and other insects were controlled with orizon (Producer, location of producer) (active ingredient, acetamiprid, and abamectin) using 0.133% v/v. Fungi were controlled using ridomil gold (Producer, Location of producer) at 2% w/v.Determination of growth and yield of tomato and cucumberThree tomato and cucumber plants were randomly selected for each treatment for the determination of growth parameters (plant height, leaf area, number of leaves per plant, and stem diameter) at mid the flowering stage of tomato and cucumber plants.The leaf area of tomato was calculated using the model (A = KL2) developed by Lyon30, where L = Length of tomato leaf, K = constant which is 0.1551, and A = leaf area of tomato. Similarly, the leaf area of cucumber was calculated using A = 0.88LW – 4.27, where L = cucumber leaf length and W = cucumber leaf width, A = leaf area of cucumber31.Tomato fruits were ready for harvest from 65 days after transplanting, harvestings were done twice every week (Mondays and Fridays) for up to 85 days after transplanting. Similarly, harvesting of cucumber fruits started 35 days after transplanting and harvestings were also done twice a week (Mondays and Fridays), harvesting was carried out till 60 days after transplanting. Tomato and cucumber fruit yields were counted and weighed at each harvest.Analysis of tomato and cucumber leaves and fruitsAt the 50% flowering stage of tomato and cucumber plants, ten leaf samples were collected from each treatment. The leaf samples were oven-dried at 75 °C for 24 h and thereafter grounded. The grounded samples were later analyzed for nitrogen (N), phosphorous (P), potassium (K), calcium (Ca), and magnesium (Mg) content using the method of described by32. At harvest, four matured tomato and cucumber fruits of uniform size were selected per treatment, and their nutrient compositions were determined using the method of33.Statistical analysisAll data collected on the growth, yield, leaf, and fruit nutrient contents of tomato and cucumber were subjected to analysis of variance (ANOVA). The SPSS V 21.0 (New York, USA) software was used to perform ANOVA and Duncan’s multiple range test (DMRT) was used to compare means at a 5% probability level.
Ethical approvalI confirm that all the research meets ethical guidelines and adheres to the legal requirements of the study country.Compliance with international, national and/or institutional guidelinesExperimental research (either cultivated or wild), comply with relevant institutional, national, and international guidelines and legislation. Experimental studies were carried out in accordance with relevant institutional, national or international guidelines or regulation. More

Gifford-Gonzalez, D. Bones are not enough: Analogues, knowledge, and interpretive strategies in zooarchaeology. J. Anthropol. Archaeol. 10, 215–254 (1991).Article 

Google Scholar 
Pobiner, B. L. The zooarchaeology and paleoecology of early hominin scavenging. Evol. Anthropol. 2, 68–82 (2020).Article 

Google Scholar 
Rodriguez, A. et al. Right or left? Determining the hand holding the tool from use traces. J. Archaeol. Sci. Rep. 31, 102316 (2020).
Google Scholar 
Feix, T., Kivell, T. L., Pouydebat, E. & Dollar, A. M. Estimating thumb-index finger precision grip and manipulation potential in extant and fossil primates. J. R. Soc. Interface. https://doi.org/10.1098/rsif.2015.0176 (2015).Article 
PubMed 
PubMed Central 

Google Scholar 
Bardo, A. et al. The implications of thumb movements for Neanderthal and modern human manipulation. Sci. Rep. 10, 19323 (2020).ADS 
CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Stout, D., Semaw, S., Rogers, M. J. & Cauche, D. Technological variation in the earliest Oldowan from Gona, Afar, Ethiopia. J. Hum. Evol. 58, 474–491 (2010).PubMed 
Article 

Google Scholar 
Tennie, C., Premo, L. S., Braun, D. R. & McPherron, S. P. Resetting the null hypothesis: Early stone tools and cultural transmission. Curr. Anthrop. 58, 652–672 (2017).Article 

Google Scholar 
Tennie, C. The zone of latent solution (ZLS) account remains the most parsimonious explanation for early stone tools. Curr. Anthrop. 60, 331–332 (2019).
Google Scholar 
Tennie, C., Braun, D. R., Premo, L. S. & McPherron, S. P. The Island Test for Cumulative Culture in Paleolithic Cultures. In The Nature of Culture. Series: Vertebrate Paleobiology and Paleoanthropology (eds Haidle, M. N. et al.) (Springer, 2016).
Google Scholar 
Perreault, C. The Quality of the Archaeological Record (University of Chicago Press, 2019).Book 

Google Scholar 
Proffitt, T. et al. Wild monkeys flake stone tools. Nature 539, 85–88 (2016).ADS 
CAS 
PubMed 
Article 

Google Scholar 
Carvalho, S., Cunha, E., Sousa, C. & Matsuzawa, T. Chaînes opératoires and resource-exploitation strategies in chimpanzee (Pan troglodytes) nut cracking. J. Hum. Evol. 55, 148–163 (2008).PubMed 
Article 

Google Scholar 
Westergaard, G. C. & Suomi, S. J. The stone tools of capuchins (Cebus apella). Int. J. Primatol. 16, 1017–1024 (1995).Article 

Google Scholar 
De la Torre, I. & Mora, R. Technological Strategies in the Lower Pleistocene at Olduvai Beds I and II (Service de Prehistoire, Universite de Liege, 2005).
Google Scholar 
M. D. O. M. Í. Dominguez-Rodrigo, 3.3-Million-Year-Old Stone Tools and Butchery Traces? More Evidence Needed. PaleoAnthropology. 9 (2016).Harmand, S. et al. 3.3-million-year-old stone tools from Lomekwi 3, West Turkana, Kenya. Nature 521, 310–315 (2015).ADS 
CAS 
PubMed 
Article 

Google Scholar 
Andrefsky, W. Lithics: Macroscopic Approaches to Analysis (Cambridge University Press, 2005).Book 

Google Scholar 
Malaivijitnond, S. et al. Stone-tool usage by Thai long-tailed macaques (Macaca fascicularis). Am. J. Primatol. 69, 227–233 (2007).PubMed 
Article 

Google Scholar 
Luncz, L. V. et al. Resource depletion through primate stone technology. eLife 6, e23647 (2017).PubMed 
PubMed Central 
Article 

Google Scholar 
Leca, J. B., Gunst, N. & Huffman, M. Complexity in object manipulation by Japanese macaques (Macaca fuscata): A cross-sectional analysis of manual coordination in stone handling patterns. J. Comp. Psychol. 125, 61 (2011).PubMed 
Article 

Google Scholar 
Toth, N., Schick, K. D., Savage-Rumbaugh, E. S., Sevcik, R. A. & Rumbaugh, D. M. Pan the tool-maker: Investigations into the stone tool-making and tool-using capabilities of a bonobo (Pan paniscus). J. Archaeol. Sci. 20, 81–91 (1993).Article 

Google Scholar 
Wright, R. V. S. Imitative learning of a flaked stone technology-The case of an orangutan. Mankind 8, 296–306 (2009).
Google Scholar 
Bandini, E. et al. Naïve, unenculturated chimpanzees fail to make and use flaked stone tools. Open Res. Eur. https://doi.org/10.12688/openreseurope.13186.2 (2021).Article 
PubMed 
PubMed Central 

Google Scholar 
J. Henrich, C. Tennie, in Chimpanzees and Human Evolution, M. Muller, R. Wrangham, D. Pilbeam, Eds. (Harvard University Press, Cambridge, MA, (2017), 645–702.Breuer, T., Ndoundou-Hockemba, M. & Fishlock, V. First observation of tool use in wild gorillas. PLoS Biol. 3, 2041–2043 (2005).CAS 
Article 

Google Scholar 
Wittiger, L., Society, W. C., River, C. & Project, G. Tool use during display behavior in wild cross river gorillas. Am. J. Primat. 5, 1–5 (2007).
Google Scholar 
Kinani, J. F. & Zimmerman, D. Tool use for food acquisition in a wild mountain gorilla (Gorilla beringei beringei). Am. J. Primat. 77, 353–357 (2015).Article 

Google Scholar 
Grueter, C. C., Robbins, M. M., Ndagijimana, F. & Stoinski, T. S. Possible tool use in a mountain gorilla. Behav. Processes. 100, 160–162 (2013).PubMed 
Article 

Google Scholar 
Parker, S. T., Kerr, M., Markowitz, H. & Gould, J. A survey of tool use in zoo gorillas. In The Mentalities of Gorillas and Orangutans: Comparative Perspectives (eds Parker, S. T. et al.) (Cambridge University Press, 1999).Chapter 

Google Scholar 
Shumaker, R. W., Walkup, K. R. & Beck, B. B. Animal Tool Behavior: The Use and Manufacture of Tools by Animals (Johns Hopkins University Press, 2011).
Google Scholar 
Pouydebat, E., Berge, C., Gorce, P. & Coppens, Y. Use and manufacture of tools to extract food by captive Gorilla gorilla gorilla: Experimental approach. Folia Primat. 76, 180–183. https://doi.org/10.1159/000084381 (2005).Article 

Google Scholar 
Haslam, M. ‘Captivity bias’ in animal tool use and its implications for the evolution of hominin technology. PTRBAE 368, 20120421 (2013).PubMed 
PubMed Central 

Google Scholar 
Van Schaik, C. P., Deaner, R. O. & Merrill, M. Y. The conditions for tool use in primates: Implications for the evolution of material culture. J. Hum. Evol. 36, 719–741 (1999).PubMed 
Article 

Google Scholar 
Pouydebat, E., Gorce, P., Coppens, Y. & Bels, V. Biomechanical study of grasping according to the volume of the object: Human versus non-human primates. J. Biomech. 42, 266–272 (2009).PubMed 
Article 

Google Scholar 
Pouydebat, E., Laurin, M., Gorce, P. & Bels, V. Evolution of grasping among anthropoids. J. Evol. Bio. 21, 1732–1743 (2008).CAS 
Article 

Google Scholar 
Bardo, A., Cornette, R., Borel, A. & Pouydebat, E. Manual function and performance in humans, gorillas and orangutans during the same tool use task. Am. J. Phys. Anthropol. https://doi.org/10.1002/ajpa.2332 (2017).Article 
PubMed 

Google Scholar 
A. Bardo, A. Borel, H. Meunier, J. P. Guéry, E. Pouydebat, Manual abilities in great apes during a tool use task. Am. J. Phys. Anthropol. doi: 10.1002 (2016).W. C. McGrew, Why is ape tool use so confusing. Comparative socioecology: the behavioural ecology of humans and other mammals. 457–472 (1989).Cipolletta, C. et al. Termite feeding by Gorilla gorilla gorilla at Bai Hokou, Central African Republic. Int. J. Primatol. 28, 457–476 (2007).Article 

Google Scholar 
Salmi, R., Rahman, U. & Doran-Sheehy, D. M. Hand preference for a novel bimanual coordinated task during termite feeding in wild western gorillas (Gorilla gorilla gorilla). Int. J. Primatol. 37, 200–212 (2016).Article 

Google Scholar 
Masi, S. et al. The influence of seasonal frugivory on nutrient and energy intake in wild western gorillas. PLoS ONE 10, e0129254 (2015).PubMed 
PubMed Central 
Article 
CAS 

Google Scholar 
Redford, K. H. & Dorea, J. G. The nutritional value of invertebrates with emphasis on ants and termites as food for mammals. J. Zool. 203, 385–395 (1984).CAS 
Article 

Google Scholar 
McGrew, W. C. The ‘other faunivory’revisited: Insectivory in human and non-human primates and the evolution of human diet. J. Hum. Evol. 71, 4–11 (2014).PubMed 
Article 

Google Scholar 
Tennie, C., O’Malley, R. C. & Gilby, I. C. Why do chimpanzees hunt? Considering the benefits and costs of acquiring and consuming vertebrate versus invertebrate prey. J. Hum. Evol. 71, 38–45 (2014).PubMed 
Article 

Google Scholar 
McBrearty, S. Consider the humble termite: Termites as agents of post-depositional disturbance at African archaeological sites. J. Archaeol. Sci. 17, 111–143 (1990).Article 

Google Scholar 
Okwakol, M. J. N. Effects of Cubitermes testaceus (Williams) on some physical and chemical properties of soil in a grassland area of Uganda. Afr. J. Ecol. 25, 147–153 (1987).Article 

Google Scholar 
Altmann, J. Observational study of behavior: Sampling methods. Behavior 49, 227–267 (1974).CAS 
Article 

Google Scholar 
Robira, B. et al. Handedness in gestural and manipulative actions in male hunter-gatherer Aka pygmies from Central African Republic. Am. J. Phys. Anthropol. 166(481–491), 19 (2018).
Google Scholar 
Meguerditchian, A., Calcutt, S. E., Lonsdorf, E. V., Ross, S. R. & Hopkins, W. D. Brief communication: Captive gorillas are right-handed for bimanual feeding. Am. J. Phys. Anthropol. 141, 638–645 (2010).PubMed 
PubMed Central 

Google Scholar 
Dapena, J. E. S. Ú. S., William, J., Anderst, N. P. & Toth, The biomechanics of the arm swing in Oldowan stone flaking. In The Oldowan: Case Studies into the Earliest Stone Age (No. 1). Gosport (eds Toth, N. P. & Schick, K. D.) (Stone Age Institute Press, 2006).
Google Scholar 
Nowell, A. A. & Fletcher, A. W. The development of feeding behaviour in wild western lowland gorillas (Gorilla gorilla gorilla). Behaviour 145, 171–193 (2008).Article 

Google Scholar 
Pouydebat, E., Gorce, P., Coppens, Y. & Bels, V. Substrate optimization in nuts cracking by capuchin monkeys. Am. J. Primatol. 68, 1017–1024 (2006).CAS 
PubMed 
Article 

Google Scholar 
Boinski, S., Quatrone, R. P. & Swarttz, H. Substrate and tool use by brown capuchins in Suriname: Ecological contexts and cognitive bases. Am. Anthropol. 102, 741–761 (2000).Article 

Google Scholar 
Panger, M. A. Object-use in free-ranging white-faced capuchins (Cebus capucinus) in Costa Rica. Am. J. Phys. Anthropol. 106, 311–321 (1998).CAS 
PubMed 
Article 

Google Scholar 
Parker, S. T. & Gibson, K. R. Object manipulation, tool use and sensorimotor intelligence as feeding adaptations in cebus monkeys and great apes. J. Hum. Evol. 6, 623–641 (1977).Article 

Google Scholar 
Heldstab, S. A., Isler, K., Schuppli, C. & van Schaik, C. P. When ontogeny recapitulates phylogeny: Fixed neurodevelopmental sequence of manipulative skills among primates. Sci. Adv. 6, eabb4685 (2020).ADS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Clutton-Brock, T. H. Some aspects of intraspecific variation in feeding and ranging behaviour in primates. In Primate Ecology Studies of Feeding And Ranging Behavior in Lemurs, Monkeys and Apes (ed. Clutton-Brock, T. H.) (Academic Press, 1977).
Google Scholar 
Key, C. & Ross, C. Sex differences in energy expenditure in non-human primates. Proc. R. Soc. Lond. B. 266, 2479–2485 (1999).CAS 
Article 

Google Scholar 
Lockman, J. J. A perception–action perspective on tool use development. Child Dev. 71, 137–144 (2000).CAS 
PubMed 
Article 

Google Scholar 
Masi, S. et al. Unusual feeding behavior in wild great apes, a window to understand origins of self-medication in humans: Role of sociality and physiology on learning process. Physiol. Behav. 105, 337–349 (2012).CAS 
PubMed 
Article 

Google Scholar 
Falótico, T. & Ottoni, E. B. The manifold use of pounding stone tools by wild capuchin monkeys of Serra da Capivara National Park, Brazil. Behaviour 153, 421–442 (2016).Article 

Google Scholar 
Falótico, T. & Ottoni, E. B. Stone throwing as a sexual display in wild female bearded capuchin monkeys, Sapajus libidinosus. PLoS ONE 8, e79535 (2013).ADS 
PubMed 
PubMed Central 
Article 
CAS 

Google Scholar 
Mannu, M. & Ottoni, E. B. The enhanced tool-kit of two groups of wild bearded capuchin monkeys in the Caatinga: Tool making, associative use, and secondary tools. Am. J. Primatol. 71, 242–251 (2009).PubMed 
Article 

Google Scholar 
Gumert, M. D., Kluck, M. & Malaivijitnond, S. Thephysical characteristics and usage patterns of stoneaxe and pounding hammers used by long-tailedmacaques in the Andaman Sea region of Thailand. Am. J. Primatol. 71, 594–608. https://doi.org/10.1002/ajp.20694 (2009).Article 
PubMed 

Google Scholar 
Marzke, M. W. Precision grips, hand morphology, and tools. Am. J. Phys. Anthropol. 102, 91–110 (1997).CAS 
PubMed 
Article 

Google Scholar 
Matsuzawa, T. Chimpanzee Intelligence in Nature and in Captivity Isomorphism of Symbol Use and Tool Use (Cambridge University Press, 1996).Book 

Google Scholar 
Westergaard, G. C. & Suomi, S. J. A simple stone-tool technology in monkeys. J. Hum. Evol. 27, 399–404 (1994).Article 

Google Scholar 
Liu, Q. et al. Kinematics and energetics of nut-cracking in wild capuchin monkeys (Cebus libidinosus) in Piauí, Brazil. Am. J. Phys. Anthropol. 138, 210–220 (2009).CAS 
PubMed 
Article 

Google Scholar 
Günther, M. M. & Boesch, C. Energetic Cost of Nut-cracking Behaviour in Wild Chimpanzees. In Hands of Primates 109–129 (Springer, 1993).
Google Scholar 
Roach, N. T., Venkadesan, M., Rainbow, M. J. & Lieberman, D. E. Elastic energy storage in the shoulder and the evolution of high-speed throwing in Homo. Nature 498, 483–486 (2013).ADS 
CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Young, N. M., Capellini, T. D., Roach, N. T. & Alemseged, Z. Fossil hominin shoulders support an African ape-like last common ancestor of humans and chimpanzees. PNAS 112, 11829–11834 (2015).ADS 
CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Doran-Sheehy, D., Mongo, P., Lodwick, J. & Conklin-Brittain, N. L. Male and female western gorilla diet: Preferred foods, use of fallback resources, and implications for ape versus old world monkey foraging strategies. Am. J. Phys. Anthropol. 140, 727–738 (2009).CAS 
PubMed 
Article 

Google Scholar 
Breuer, T., Hockemba, M. B. N., Olejniczak, C., Parnell, R. J. & Stokes, E. J. Physical maturation, life-history classes and age estimates of free-ranging western gorillas – Insights from Mbeli Bai, Republic of Congo. Am. J. Primatol. 71, 106–119 (2009).PubMed 
Article 

Google Scholar 
Hopkins, W. D. et al. The use of bouts and frequencies in the evaluation of hand preferences for a coordinated bimanual task in chimpanzees (Pan troglodytes): An empirical study comparing two different indices of laterality. J. Comp. Psychol. 115, 294–299 (2001).CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Byrne, R. W. & Byrne, J. M. Manual dexterity in the gorilla: bimanual and digit role differentiation in a natural task. Anim. Cogn. 4, 347–361 (2001).CAS 
PubMed 
Article 

Google Scholar  More