Available online at www.sciencedirect.com An integrated eco-evolutionary framework to predict population-level responses of climate-sensitive pathogens Amy M Campbell 1,2, Chris Hauton 1, Craig Baker-Austin 2, ]] ]]]]]] ]] Ronny van Aerle 2 and Jaime Martinez-Urtaza 2,3 It is critical to gain insight into how climate change impacts exposure to extreme weather events [1,2]. Additionally, evolutionary responses within climate-sensitive pathogen these changing environmental conditions are altering the populations, such as increased resilience, opportunistic epidemiology of diseases caused by climate-sensitive responses and the emergence of dominant variants from pathogens that exist in natural ecosystems [1,3], wherein highly variable genomic backgrounds and subsequent global pathogen life cycles are intrinsically linked to environ­ dispersal. This review proposes a framework to support such mental conditions (e.g. Vibrio parahaemolyticus, a water- analysis, by combining genomic evolutionary analysis with borne bacteria causing gastroenteritis [4]), or their host climate time-series data in a novel spatiotemporal dataframe vector is governed itself by the environment (e.g. Borrelia for use within machine learning applications, to understand burgdorferi, the tick-borne bacteria causing Lyme disease past and future evolutionary pathogen responses to climate [5,6]). Known mechanisms of how diseases are driven by change. Recommendations are presented to increase the climate change include seasonal shifts and expansions of feasibility of interdisciplinary applications, including the transmission periods [1,3], and spatial range expansions importance of robust spatiotemporal metadata of both vectors and pathogens themselves (transconti­ accompanying genome submission to databases. Such nental, latitudinal and polewards) [4–10]. However, re­ workflows will inform accessible public health tools and cent years have seen the transcontinental expansion of early-warning systems, to aid decision-making and mitigate certain pathogen variants with pandemic potential [9,10]. future human health threats. The mechanisms of such emergence are unknown, as little research has moved beyond well-established eco­ Addresses logical relationships to quantify potential evolutionary 1 School of Ocean and Earth Science, University of Southampton, responses of the pathogen to changes in their native National Oceanography Centre, Southampton, UK 2 environment. Centre for Environment, Fisheries and Aquaculture Science (CEFAS), Weymouth, UK 3 Department of Genetics and Microbiology, Autonomous University of Specifically, there is a need to characterise how evolu­ Barcelona, Barcelona, Spain tionary drivers of individual strains or clonal groups (such as mutation and horizontal gene transfer) and demo­ Corresponding author: Martinez-Urtaza, Jaime (jaime.martinez.urtaza@uab.cat) graphy can be modulated by climate change. Environmental variability, such as temperature changes, leads to pathogen variants either thriving in the new Current Opinion in Biotechnology 2023, 80:102898 conditions, due to phenotypic plasticity, or adapting [11], This review comes from a themed issue on Environmental indicated by the presence of candidate genes under Biotechnology positive selection [12] that correlate with particular bio­ Edited by Jaime Martinez Urtaza and Luis F. De León climatic conditions [13], which can lead to clonal re­ For complete overview of the section, please refer to the article placement by well-adapted variants. Such adaptation collection, “Environmental Biotechnology (2023)” could offer a range of benefits, including enhanced sur­ Available online 3 February 2023 vivability, fitness or transport potential. https://doi.org/10.1016/j.copbio.2023.102898 There is currently no climate-sensitive disease model­ 0958–1669/© 2023 The Author(s). Published by Elsevier Ltd. This is ling tool that considers a genetic component [14]. The an open access article under the CC BY-NC-ND license (http:// next generation of tools will need to evaluate environ­ creativecommons.org/licenses/by-nc-nd/4.0/). mental drivers against observed demographic events (e.g. emergence, bottlenecks, population expansions and migrations) or the uptake of particular genetic material, to retrospectively identify selection pressures introduced by climate change [15], and identify subsequent real- Introduction world responses of pathogen populations — which have Climate change has implications for human health, in previously only been hypothesised [6,16] or explored in terms of heat vulnerability, food and water insecurity and lab-based studies of adaption and tolerances [17,18]. www.sciencedirect.com Current Opinion in Biotechnology 2023, 80:102898 2 Environmental Biotechnology Box 1 Limitations of existing, individual ecological and evolutionary approaches Existing ecological models that have been used to explore how species are responding to climate change, have certain limitations when applied to capture the full dynamics of complex evolutionary responses of climate-sensitive pathogens. Often, they assume the species to be an unchanging entity, through optimum niche habitat shifts and dispersal based on physiological limitations of the species, without considering any evolutionary component or opportunities for adaption. Models are based on explicit occurrence/abundance population data [11] and do not allow the in­ troduction of genomic data. Additionally, recently studies exploring evolutionary responses to climate change in terms of population survivability [13] or heterogeneous intra-species responses [38,43], are developed for larger, slower-evolving taxa, such as plants or animals [38,44,45]. Pathogens, however, exhibit rapid evolutionary change on faster timescales with significant horizontal gene transfer, requiring adapted workflows as those that exist are not directly applicable. Finally, certain ecological models act at a species level — but pathogens may not respond homogeneously at a species level, as particularly well-adapted clonal groups emerge from highly variable genomic backgrounds, with multiple variants coexisting in a population, requiring the exploration of individual variant responses and population dynamics. We explore some examples of ecological models used for similar applications below: • Species Distribution Models — also known as ecological niche modelling, the niche of a species is estimated and mapped based on statistical correlations between species occurrence and environmental variables, mostly using Maximum Entropy or Generalised Linear Models [46]. These are known to underestimate [47] or overestimate [48] range boundaries based on simplistic assumptions that ignore adaption or geographic barriers. Evolutionary aspects are omitted — species are treated as having uniform requirements, which is inconsistent with our knowledge of heterogeneous groups of individuals existing in a constant state of change within a single pathogenic species [9]. • Hybrid Genomic Species Distribution Models — which instead of species occurrence, can focus on a particular genomic process, for example, identifying specific SNP niches, through geographical correlation between specific SNP frequencies and environmental variables, for ex­ ample, [13,43–45,49]. Recent studies have been able to add an adaptive evolutionary component in this way, but this still requires combi­ nation with further genomic analysis to provide further evolutionary or demographic information, that is, migration rates, evolutionary rates and effective population size [11] to understand population-level responses, with limited methodologies available to do so. Studies are geared towards genetic vulnerability studies of animal/plant species based on discrepancies in allele presence (assuming current populations are well-adapted to local environments) — whereas we need to understand opportunistic pathogen evolution that can rapidly respond to such changes. • Metapopulation Models — these aim to project demographic consequences of populations to changing environmental variables, allowing disconnected patches of species to have independent dynamics and dispersal connections [50]. However, the demography of pathogen populations is largely more complex and dynamic than those of larger taxa the models are developed for, with highly variable genomic backgrounds of many more distinct genotypes, and high rates of horizontal gene transfer between these, upon which certain well-adapted ones emerge — making it difficult to redeploy such models for pathogen evolution. Understanding connections between populations would require the explicit integration of phylogenetic analysis within the model. Similarly, genomic approaches that aim to quantify how species respond to climate change are explored by Waldvogel et al. [11] who only elu­ cidate a subset of evolutionary processes rather than fully integrating eco-evolutionary dynamics [11]. New approaches, specific to pathogens, have emerged recently, for example, detecting bacterial DNA transfer providing adaption to new environments [51], but require a spatiotemporal element to facilitate integration with environmental data to explicitly pinpoint such selective pressures and responses. The gap between ecological models and such genomic approaches needs to be bridged with an infrastructure that facilitates seamless integration of eco-evolutionary processes. This review proposes an option to overcome these lim­ into optimum range shifts and geographic abundance, itations through a framework able to integrate genomic evolutionary genomics, in the form of genomic and data, satellite-derived climate data, epidemiological data phenotypic data, are necessary to provide population- and pre-existing tools within machine learning applica­ level insights and predictions into genetic adaption, tions to provide novel, interdisciplinary insights into phenotypic plasticity and demography shifts in response current and future evolutionary pathogen responses to to climate change [11]. climate change. This framework will allow us to develop real-time public health tools and early-warning systems that can anticipate the future climate-driven emergence An eco-evolutionary framework of well-adapted pathogenic strains and mitigate future The framework outlined here (Figure 1) proposes that pandemics, ultimately increasing the resilience of health various pre-existing software are redeployed to integrate systems to climate-related risks [19]. ecological dynamics and evolutionary processes, in the context of climate-sensitive pathogens. The seamless integration of phylogenetic analysis, climate timeseries, Beyond ecology and additional data or encoders into a novel dataframe Existing ecological models, that aim to quantify how will facilitate explicit statistical investigation of complex species will respond to climate change, are not sufficient eco-evolutionary relationships between climate change to integrate genomic data and capture the full dynamics and pathogens. The dataframe will provide training data of pathogen-specific evolutionary responses to climate for machine learning applications to facilitate forecasting change (Box 1). The advent of eco-evolutionary models of future pathogen epidemiology, for example, optimum is necessary — while ecological data can give insights evolution zones and population-specific forecasts. Near- Current Opinion in Biotechnology 2023, 80:102898 www.sciencedirect.com Eco-evolutionary framework to predict pathogen Campbell et al. 3 Figure 1 Current Opinion in Biotechnology An eco-evolutionary framework to explore changing epidemiology of climate-sensitive pathogens. A range of interdisciplinary inputs are explored, which are then integrated in a spatiotemporal dataframe, to be used as inputs in machine learning applications. Outputs will be implemented in integrated tools for climate-sensitive pathogen surveillance and forecasts. Created with BioRender.com. www.sciencedirect.com Current Opinion in Biotechnology 2023, 80:102898 4 Environmental Biotechnology Box 2 Spatiotemporal data requirements and analysis limitations The toolkit requires genomes (tree tips) to be georeferenced with latitude and longitude values, and time-stamped in the metadata to allow inference of the spatiotemporal context of phylogenetic tree branches and internal tree nodes [15]. Genomic metadata on public repositories is currently insufficient. Here, we strongly recommend the ‘Latitude/Longitude’, ‘Location’ and time-stamped ‘Collection Date’ (in a ‘dd-mm-yyyy’ format) fields are completed during genomic data submission to databases. A range of software currently exists that can visualise spatiotemporal relationships by projecting phylogenetic trees in geographical space [23,24,52–54], however, they cannot explicitly facilitate spatiotemporal correlation analysis of driver variables in-house. These inferences would be strengthened with integrated workflows that allow hypotheses to be tested [15] — currently this is only possible for static, univariate correlation analysis between environmental variables and allele frequencies [55] or phylogeny branches [24], omitting complex multivariate relationships between the constantly evolving pathogen and its dynamic environment. real-time surveillance and climate data will be con­ and supported by robust surveillance systems, are also tinuously re-fed into models for reanalysis and refine­ useful for cross-validation. ment. Such a template will be invaluable in the field of epidemiology and human health, but is flexible enough to be applied to a range of climate-sensitive species to Analysis increase our understanding of contemporary biodiversity Genome-wide analysis to extract evolutionary events and extinction processes amidst climate change [15]. Bayesian phylogenetic analysis of molecular sequences is utilised to reconstruct phylogenies to pinpoint the Inputs time and location of evolutionary events and demo­ Climate data graphic shifts, to explore their associations with climate Spatiotemporal data of a range of environmental vari­ anomalies, requiring dated tips with discrete (country ables affecting environmental pathogens are required for or region) or continuous location data (co-ordinates). retrospective analysis, to identify conditions that drive Phylogenetic outputs (maximum clade credibility evolutionary shifts. This requires high-resolution data trees) are converted into more functional outputs (e.g. (as pathogen niches are specific and short-lived due to dataframes, KML) for further spatiotemporal analysis rapid replication) and extensive coverage (to elucidate using various tools [23,24]. Demographic shifts in introduction and transport mechanisms [20]). While field dominant sequence types and population structure are data collection at this scale is unfeasible, solutions can be identified using multilocus sequence typing along a found in the field of remote sensing that utilises con­ timeseries or clusters of single-nucleotide polymorph­ tinuous satellite observations to provide consistent, isms (SNPs). Genome-wide association studies spatial coverage of climate variables for analysis, at up to (GWAS) identify the presence of genes with evolu­ a 1 km and hourly resolution [21]. Finally, climate fore­ tionary or adaptive significance, to elucidate popula­ cast data and projections will become increasingly im­ tion-specific responses used by pathogens to thrive and portant towards the end of the workflow to predict future establish themselves in geographically distinct, dy­ eco-evolutionary pathways for pathogens once these re­ namic environments. Annotating the trend and dis­ lationships have been established. tribution of accessory gene presence reveals selective signatures and horizontally transferred genetic mate­ Genomic data rial — with genes under positive selection [12], sug­ Tools need to integrate analysis of evolutionary and gesting a benefit that outweighs the pace of genetic demographic timescales against timeseries of climate drift [25]. Recombination events and rates are identi­ data to interrogate possible interactions and anticipate fied using various software [26–29]. These ‘evolu­ future evolutionary pathways. To facilitate this analysis, tionary metrics’ are all converted into useable formats robust genomic surveillance is necessary — ideally pro­ in the next stage for downstream analyses. viding a wide range of sequenced pathogen genomes from which to infer evolutionary processes. The chal­ lenging spatiotemporal requirements of genomic sur­ Spatiotemporal dataframe generation veillance for this application are discussed in Box 2. Existing spatiotemporal software to explore climate- driven evolution is unable to identify explicit, complex Additional input datasets associations (Box 2). This framework advocates for the Additional datasets to obtain, dependent on user re­ generation of a fit-for-purpose spatiotemporal dataframe, quirements, include human demographic data (shipping similar to that used in Campbell et al. [30] but featuring and transport routes, exposure), for novel analysis into evolutionary-dependent variables, to deploy in various the global dynamics of pathogens amidst human mobi­ statistical applications and provide appropriate training lity [22], and epidemiological data, which, if available data for machine learning applications. Current Opinion in Biotechnology 2023, 80:102898 www.sciencedirect.com Eco-evolutionary framework to predict pathogen Campbell et al. 5 Each data point in the nested spatiotemporal dataframe accommodating for heterogeneous and complex pa­ is aggregated by a time unit, such as month, and spatial thogen dynamics. ML has been utilised to forecast po­ unit, such as district. Evolutionary metrics and gene pulation-level heat tolerances [34–37], environmental presence are converted into a binary format (attributed drivers of adaption [38] and diffusion [33] in slower- a 0 or 1), known as ‘one hot encoding’, to indicate if evolving taxa, but it has not been fully exploited to events were observed in this particular spatiotemporal elucidate the evolutionary responses of pathogens to unit. Each step is then appended with associated re­ climate change, likely due to the absence of a toolkit to levant climate data and appropriate lagged or en­ combine genomic and climate datasets into a cohesive gineered features, using a zonal extraction function (as format for predictive applications [34], which our fra­ used in Ref. [30]), alongside any further discrete or mework aims to solve. continuous ancillary data, for example, socioeconomic data. This seamless integration of environmental and Specifically, ML tools will be developed on the pro­ genomic data offers analysis potential for a range of posed novel spatiotemporal dataframe (Figure 2). The evolutionary components against climate data in com­ spatiotemporal dataframe will be split 80%-10%-10% (or bination, which is lacking in pre-existing models similar) into a training dataset, validation dataset (for (Box 1). model refinement) and unseen test dataset (to draw ac­ curacy metrics from), respectively. The ML model re­ Statistical multivariate correlation analysis sults can be interrogated to extract feature importance Multivariate regression (using models such as ordinary and infer the relative contribution of each climate vari­ least squares, principal component analysis and gen­ able to the final predictions. Trained and refined ML eralised additive models) can expose statistically sig­ models can then be applied to long-term or real-time nificant relationships between the environmental forecast data (or hypothetical future climate scenarios) to variables and evolutionary metrics. Nonlinear associa­ anticipate a range of evolutionary events — including tions can be revealed using gradient boosting trees the likelihood of mutations, gene transfer or presence of (which can reveal the rank and relative importance of particular adaptive traits giving an indication on the fu­ each variable in driving the evolutionary metric) and ture global ecology and evolution of a range of patho­ partial dependence plots to show the marginal effect of gens. Epidemiological and surveillance data will be key each variable on the evolutionary metric of interest. to validating the trained models, by using real-world Significant relationships are those within a 95% con­ outbreak data to give indications of forecast performance fidence interval (p = < 0.05). The dataframe can be split [30], and for reanalysis to provide continuous model re­ into spatially or temporally distinct subgroups to explore finement. spatial or seasonal variations in correlation coefficients. However, the majority of these relationships will likely be complex, and so statistical analysis alone may struggle Looking forward: output opportunities and to explore thresholds and limiting factors of variables limitations interacting in a multivariate landscape. The outputs of the framework can then be used for various resulting applications, within academia or policy Machine learning applications for forecasting and decision-making — particularly the scientific results Evolutionary biology is extending beyond only re­ should be converted into user-friendly public health constructing the past, towards harnessing this to predict tools. The accuracy of these outputs is dependent on the future evolutionary processes and dynamics for in­ input datasets, which will therefore benefit from in­ dividual populations [31]. Implementing this requires an creased data quality, availability and standardisation. It is effective method of integrating a complex range of in­ critical to fund, develop and maintain active surveillance terdisciplinary data, spanning all aspects of climate eco- systems in the future, to provide an abundance of gen­ evolutionary analysis, to make forecasts — a complex omes from which to identify new epidemic clones and task that has proved difficult in previous studies [32,33]. pinpoint when and how they emerge successfully from a population comprising many variants, with the recent Machine learning (ML), in which the system can ‘learn’ Covid-19 pandemic giving an insight into what can be relationships between components of the model, is a achieved with high-quality, near-real-time genomic se­ computationally feasible way to effectively synthesise quencing to identify the location of emerging strains and climate and genomic datasets and draw meaningful consequent outbreaks [39,40]. Data are often scattered predictions from heterogeneous, population-specific across a range of community-specific repositories or nonlinear interactions, functioning on a complicated paper supplements [41,42] making interdisciplinary re­ spatiotemporal, multivariate landscape. ML offers a search difficult and requiring fit-for-purpose inter­ range of modelling options itself, or has the ability to disciplinary databases for open data sharing and enhance the potential of existing ecological models by conciliation of results of climate-sensitive pathogens or overcoming the limitations mentioned in Box 1, other species [41]. www.sciencedirect.com Current Opinion in Biotechnology 2023, 80:102898 6 Environmental Biotechnology Figure 2 Current Opinion in Biotechnology Implementation of Machine Learning within the eco-evolutionary framework, highlighting merits and considerations of the analysis. (TP = true positives, FP = false positives, FN = false negatives, TN = true negatives, R = recall, P = precision). Created with BioRender.com. Current Opinion in Biotechnology 2023, 80:102898 www.sciencedirect.com Eco-evolutionary framework to predict pathogen Campbell et al. 7 Concluding remarks Intergovernmental Panel on Climate Change. Cambridge This novel eco-evolutionary framework targets knowl­ University Press; 2014:1-1820. edge gaps in the analysis of how climate change is 3. Chao DL, Roose A, Roh M, Kotloff KL, Proctor JL: The seasonality of diarrheal pathogens: a retrospective study of seven sites driving climate-sensitive pathogens that exist in the over three years. PLoS Negl Trop Dis 2019, 13:e0007211, https:// environment, providing an option to overcome previous doi.org/10.1371/journal.pntd.0007211 limitations. The integration of genomic analysis in the 4. Martinez-Urtaza J, et al.: Epidemiological investigation of a next generation of tools will enhance currently well- foodborne outbreak in Spain associated with U.S. West coast genotypes of Vibrio parahaemolyticus. SpringerPlus 2016, 5:1-8, characterised ecological relationships with variant-spe­ https://doi.org/10.1186/s40064-016-1728-1 cific evolutionary responses to climate change, to un­ 5. Roy-Dufresne E, Logan T, Simon JA, Chmura GL, Millien V: derstand how strains emerge from genetically diverse Poleward expansion of the white-footed mouse (Peromyscus populations, spread to new locations and become leucopus) under climate change: implications for the spread of lyme disease. PLoS One 2013, 8:e80724, https://doi.org/10.1371/ dominant. This review advocates for genomes to be ac­ journal.pone.0080724 companied by better spatiotemporal metadata to facil­ 6. Ogden NH, Ben Beard C, Ginsberg HS, Tsao JI: Possible effects itate such analysis. ML applications have been explored of climate change on ixodid ticks and the pathogens they as facilitators to provide forecasting capabilities of future transmit: predictions and observations. J Med Entomol 2020, pathogen evolution under climate change scenarios. 58:1536-1545, https://doi.org/10.1093/jme/tjaa220 These results require follow-through to public health 7. Baker RE, et al.: Infectious disease in an era of global change. Nat Rev Microbiol (4) 2021, 20:193-205, https://doi.org/10.1038/ tools and databases to increase our understanding and s41579-021-00639-z predictive capabilities of how pathogens and other spe­ 8. Carlson CJ, et al.: Climate change increases cross-species viral cies involved in disease transmission pathways (e.g. transmission risk. Nature 2022, 607:555-562, https://doi.org/10. vectors and hosts), to ultimately reduce the human 1038/s41586-022-04788-w health burden posed by climate-sensitive pathogens. 9. Martinez-Urtaza J, et al.: Genomic variation and evolution of Vibrio parahaemolyticus st36 over the course of a transcontinental epidemic expansion. Mbio 2017, 8:e01425-17, Editorial disclosure statement https://doi.org/10.1128/mBio.01425-17 Given his role as Guest Editor, Jaime. Martinez Urtaza 10. Baker-Austin C, et al.: Vibrio spp. infections. Nat Rev Dis Prim had no involvement in the peer review of the article and 2018, 4:8, https://doi.org/10.1038/s41572-018-0005-8 has no access to information regarding its peer review. 11. Waldvogel A-M, et al.: Evolutionary genomics can improve Full responsibility for the editorial process of this article •• prediction of species’ responses to climate change. Evol Lett was delegated to Luis F. De León. 2020, 4:4-18, https://doi.org/10.1002/evl3.154. The authors discuss a range of genomic approaches that can and have been used to investigate how species will respond to climate change Data Availability and improve prediction models. Includes a discussion on the key challenges for future eco-evolutionary models. 12. Storfer A, Patton A, Fraik AK: Navigating the interface between No data were used for the research described in the article. landscape genetics and landscape genomics. Front Genet 2018, 9:68. Conflict of interest statement 13. Aguirre-Liguori JA, Ramírez-Barahona S, Gaut BS: The • evolutionary genomics of species’ responses to climate change. Nat Ecol Evol (10) 2021, 5:1350-1360, https://doi.org/10. The authors declare no conflict of interest. 1038/s41559-021-01526-9. The authors provide a methodology (accompanied by source code) to incorporate local adaption into species distribution models for a wild Acknowledgements plant species. This work was supported by the Natural Environmental Research Council [Grant number NE/S007210/1] and Centre for Environment, Fisheries and 14. Stewart Ibarra A, Diaz A, Ryan S, Lippi C, Dunbar W: Landscape Aquaculture Science (Cefas) internal Seedcorn funding. J. Martinez-Urtaza mapping of software tools for climate-sensitive infectious disease modelling. Tech. Rep., Inter-American Institute for Global is funded by grant PID2021-127107NB-I00 from Ministerio de Ciencia e Change Research and Wellcome; 2022. Innovación (Spain). 15. Fenderson LE, Kovach AI, Llamas B: Spatiotemporal landscape •• genetics: investigating ecology and evolution through space References and recommended reading and time. Mol Ecol 2020, 29:218-246. Papers of particular interest, published within the period of review, have The authors provide an overview and recommendations for analytical been highlighted as: approaches that integrate genomic and environmental data for popu­ lation structure, demography, phylogeography, metapopulation con­ •• of special interest nectivity and adaption. •• of outstanding interest 16. Casadevall A, Kontoyiannis DP, Robert V: On the emergence of Candida auris: climate change, azoles, swamps, and birds. 1. Romanello M, et al.: The 2021 report of the lancet countdown on MBio 2019, 10:e01397-19. health and climate change: code red for a healthy future. Lancet 2021, 398:1619-1662. 17. Deatherage DE, Kepner JL, Bennett AF, Lenski RE, Barrick JE: Specificity of genome evolution in experimental populations of 2. Barros VR, et al.: Climate Change 2014 Impacts, Adaptation, and Escherichia coli evolved at different temperatures. Proc Natl Vulnerability Part B: Regional Aspects: Working Group II Acad Sci USA 2017, 114:E1904-e1912, https://doi.org/10.1073/ Contribution to the Fifth Assessment Report of the pnas.1616132114 www.sciencedirect.com Current Opinion in Biotechnology 2023, 80:102898 8 Environmental Biotechnology 18. Montanchez I, et al.: Analysis of Vibrio harveyi adaptation in sea 36. Sukumaran S, Crossa J, Jarquin D, Lopes M, Reynolds MP: water microcosms at elevated temperature provides insights Genomic prediction with pedigree and genotypex environment into the putative mechanisms of its persistence and spread in interaction in spring wheat grown in South and West Asia, the time of global warming. Sci Rep 2019, 9, https://doi.org/10. North Africa, and Mexico. G3: Genes, Genomes, Genet 2017, 1038/s41598-018-36483-0 7:481-495. 19. Pörtner, H. O.et al., Climate Change 2022: Impacts, Adaptation 37. Juliana P, et al.: Integrating genomic-enabled prediction and and Vulnerability, 3056, 2022. high-throughput phenotyping in breeding for climate-resilient bread wheat. Theor Appl Genet 2019, 132:177-194. 20. Kopprio GA, et al.: Vibrio and bacterial communities across a pollution gradient in the Bay of Bengal: unraveling their 38. Layton KKS, et al.: Genomic evidence of past and future climate- biogeochemical drivers. Front Microbiol 2020, 11:594, https://doi. • linked loss in a migratory arctic fish. Nat Clim Chang 2021, org/10.3389/fmicb.2020.00594 11:158-165, https://doi.org/10.1038/s41558-020-00959-7. The authors retrospectively identify past climate-associated declines in 21. Plummer S, Lecomte P, Doherty M: The ESA Climate Change a migratory fish population using effective population size estimates and Initiative (CCI): a European contribution to the generation of the associate these with decreases in temperature. Genomic diversity and global climate observing system. Remote Sens Environ 2017, machine learning are used to identify vulnerable populations at the 203:2-8. edges of optimum temperature ranges. 22. Lemey P, et al.: Unifying viral genetics and human 39. Akther S, et al.: CoV genome tracker: tracing genomic footprints transportation data to predict the global transmission of Covid-19 pandemic. bioRxiv 2020, https://doi.org/10.1101/ dynamics of human influenza h3n2. PLoS Pathog 2014, 2020.04.10.036343 2020-04. 10:e1003932, https://doi.org/10.1371/journal.ppat.1003932 40. Forster P, Forster L, Renfrew C, Forster M: Phylogenetic network 23. Bielejec F, et al.: Spread3: interactive visualization of analysis of SARS-CoV-2 genomes. Proc Natl Acad Sci 2020, spatiotemporal history and trait evolutionary processes. Mol 117:9241-9243. Biol Evol 2016, 33:2167-2169. 41. Waldvogel A-M, Schreiber D, Pfenninger M, Feldmeyer B: Climate 24. Dellicour S, Rose R, Faria NR, Lemey P, Pybus OG: Seraphim: change genomics calls for standardized data reporting. Front studying environmental rasters and phylogenetically informed Ecol Evol 2020, 8:242, https://doi.org/10.3389/fevo.2020.00242 movements. Bioinformatics 2016, 32:3204-3206. 42. Culina A, et al.: Navigating the unfolding open data landscape in 25. Brockhurst MA, et al.: The ecology and evolution of ecology and evolution. Nat Ecol Evol 2018, 2:420-426, https://doi. pangenomes. Curr Biol 2019, 29:20. org/10.1038/s41559-017-0458-2 26. Martin DP, Murrell B, Golden M, Khoosal A, Muhire B: Rdp4: 43. Exposito-Alonso M, et al.: Genomic basis and evolutionary detection and analysis of recombination patterns in virus potential for extreme drought adaptation in Arabidopsis genomes. Virus Evol (1) 2015, 1, https://doi.org/10.1093/ve/vev003 thaliana. Nat Ecol Evol 2018, 2:352-358, https://doi.org/10.1038/ 27. Vaughan TG, et al.: Inferring ancestral recombination graphs s41559-017-0423-0 from bacterial genomic data. Genetics 2017, 205:857-870. 44. Bay RA, et al.: Genomic signals of selection predict climate- 28. Didelot X, Lawson D, Darling A, Falush D: Inference of • driven population declines in a migratory bird. Science 2018, homologous recombination in bacteria using whole-genome 359:83-86, https://doi.org/10.1126/science.aan4380. sequences. Genetics 2010, 186:1435-1449. The authors identified genes that were indicative of successful climate adaption in the yellow warbler bird, based on their associations with 29. Didelot X, Wilson DJ: Clonalframeml: efficient inference of migratory behaviour. They also found populations with the greatest recombination in whole bacterial genomes. PLoS Comput Biol shifts in allele frequencies were declining, suggesting this was due to a 2015, 11:e1004041, https://doi.org/10.1371/journal.pcbi.1004041 failure to adapt effectively. 30. Campbell AM, Racault M-F, Goult S, Laurenson A: Cholera risk: a 45. Fitzpatrick MC, Keller SR: Ecological genomics meets machine learning approach applied to essential climate • community-level modelling of biodiversity: mapping the variables. Int J Environ Res Public Health 2020, 17:9378. genomic landscape of current and future environmental adaptation. Ecol Lett 2015, 18:1-16, https://doi.org/10.1111/ele. 31. Lässig M, Mustonen V, Walczak AM: Predicting evolution. Nat 12376. Ecol Evol 2017, 1:1-9. An example of a hybrid genomic species distribution study, using a tree species as an exemplar. The authors find a particular gene that is as­ 32. Tabachnick WJ: Climate change and the arboviruses: lessons sociated with local adaption to temperature. from the evolution of the dengue and yellow fever viruses. Annu Rev Virol 2016, 3:125-145, https://doi.org/10.1146/annurev- 46. Elith J, Leathwick JR: Species distribution models: ecological virology-110615-035630 explanation and prediction across space and time. Annu Rev Ecol Evol Syst 2009, 40:677-697. 33. Jacquot M, Nomikou K, Palmarini M, Mertens P, Biek R: • Bluetongue virus spread in europe is a consequence of 47. Kearney M, Porter WP, Williams C, Ritchie S, Hoffmann AA: climatic, landscape and vertebrate host factors as revealed by Integrating biophysical models and evolutionary theory to phylogeographic inference. Proc R Soc B: Biol Sci 2017, predict climatic impacts on species’ ranges: the dengue 284:20170919. mosquito Aedes aegypti in Australia. Funct Ecol 2009, The authors fitted phylogeographic models based on environmental 23:528-538. variables to identify spatiotemporal drivers of diffusion bluetongue virus by hosts. 48. Reside AE, Critchell K, Crayn DM, Goosem M, Goosem S, Hoskin CJ, Sydes T, Vanderduys EP, Pressey RL: Beyond the model: 34. Cortés AJ, López-Hernández F, Osorio-Rodriguez D: Predicting expert knowledge improves predictions of species’ fates under • thermal adaptation by looking into populations’ genomic past. climate change. Ecol Appl 2019, 29:e01824, https://doi.org/10. Front Genet 2020, 11:564515, https://doi.org/10.3389/fgene.2020. 1002/eap.1824 564515. The authors discuss how recent thermal-driven selective responses can 49. Lowry DB, Lovell JT, Zhang L, Bonnette J, Fay PA, Mitchell RB, be inferred by coalescent methods and reconstructed using GWAS. et al.: QTL × environment interactions underlie adaptive Machine learning is then discussed for its potential to boost our pre­ divergence in switchgrass across a large latitudinal gradient. dictive capabilities in this context, informing long-term predictions of Proc Natl Acad Sci 2019, 116:12933-12941. thermal adaptive responses. 50. Jenouvrier S, Che-Castaldo J, Wolf S, Holland M, Labrousse S, 35. Garner J, et al.: Genomic selection improves heat tolerance in LaRue M, Wienecke B, Fretwell P, Barbraud C, Greenwald N, dairy cattle. Sci Rep 2016, 6:1-9. Stroeve J, Trathan PN: The call of the emperor penguin: legal Current Opinion in Biotechnology 2023, 80:102898 www.sciencedirect.com Eco-evolutionary framework to predict pathogen Campbell et al. 9 responses to species threatened by climate change. Glob 53. Argimón S, et al.: Microreact: visualizing and sharing data for Change Biol 2021, 27:5008-5029, https://doi.org/10.1111/gcb. genomic epidemiology and phylogeography. Microb Genom 15806 2016, 2:e000093. 51. Arnold BJ, Huang IT, Hanage WP: Horizontal gene transfer and 54. Parks DH, et al.: Gengis 2: geospatial analysis of traditional and adaptive evolution in bacteria. Nat Rev Microbiol 2022, genetic biodiversity, with new gradient algorithms and an 20:206-218, https://doi.org/10.1038/s41579-021-00650-4 extensible plugin framework. PLoS One 2013, 8:e69885. 52. Hadfield J, et al.: Nextstrain: real-time tracking of pathogen 55. Günther T, Coop G: Robust identification of local adaptation evolution. Bioinformatics 2018, 34:4121-4123. from allele frequencies. Genetics 2013, 195:205-220, https://doi. org/10.1534/genetics.113.152462 www.sciencedirect.com Current Opinion in Biotechnology 2023, 80:102898