Ecology

Origin of hydrate-bound hydrocarbons
A relationship between C1/(C2 + C3) and C1 δ13C has been applied to identify the sources of hydrocarbons in submarine seeps24. Recently, this diagram was revised based on a large dataset25. As shown in Fig. 4a, hydrate-bound hydrocarbons at the Kedr MV have thermogenic and/or secondary microbial origins, whereas those of other gas hydrate sites (Malenky, Bolshoy, Malyutka, Peschanka P-2, Kukuy K-0, Kukuy K-2 and Goloustnoe; Fig. 1) in Lake Baikal demonstrate microbial or early mature thermogenic origins. The hydrate-bound C1 from all locations except those at the Kedr MV were interpreted to be of microbial origin via methyl-type fermentation23 according to Whiticar’s old diagram26; however, the revised diagram25 suggests early mature thermogenic gases (Fig. 4b). Those of the Kedr MV plot at the boundary of the thermogenic and secondary microbial origin zones. Low C1 and C2 δ13C at the Peschanka P-2 MV indicated that C1 and C2 are of microbial origin27,28, whereas Kedr MV shows high C1 and C2 δ13C indicating their thermogenic origin (Fig. 4c). At other sites, C1 and C2 δ13C suggested that gases are mainly of microbial origin (in terms of C1) with some thermogenic component (13C rich and higher concentration in C2).
Stable isotopes in hydrate-bound C1 at the Kedr-1 and Kedr-2 areas suggested its thermogenic origin. However, it is close to the field of secondary microbial C1 in Fig. 4b, and the data are plotted in the overlap between the fields of thermogenic and secondary microbial in Fig. 4a. Milkov29 mentioned that secondary microbial C1 is characterised by C1-rich dry gas, large C1 δ13C (between − 55‰ and − 35‰) and large CO2 δ13C (more than + 2 ‰). Although hydrate-bound and sediment gases in the Kedr MV were not C1 rich and contained 3%–15% of C2, C1 δ13C was around − 45‰, which agrees with the secondary microbial C1. Because some data of secondary microbial gas are plotted outside the field on the original graph25, we could include the gas data in the category of secondary microbial C1 in Fig. 4b.
Figure 6 shows the relationship between C1 δ13C and CO2 δ13C in the sediment gas obtained using headspace gas method. According to the genetic diagram25, gas hydrate cores are plotted at the zones of the thermogenic and secondary microbial origins, whereas the cores at the peripheral area are primary microbial. The headspace gas data of the hydrate-bearing cores in Fig. 6 seem to be plotted in the field of thermogenic gas (low CO2 δ13C), but the effect of light CO2 produced by methane oxidation in the subsurface layer also decreased CO2 δ13C as shown in Fig. 5. These results suggested that secondary microbial C1 mixes into thermogenic gas. Coal-bearing sediments exist around the Kedr area21,22, and secondary microbial C1 can also form from coal beds30. Hydrate-bound C1 of secondary microbial origin has been only reported at the Alaska North Slope31. This study is another case for it.
Figure 6

A diagram of headspace gases. CO2 δ13C plotted against C1 δ13C, based on the classification of Milkov and Etiope25.

Full size image

Formation process of the sII gas hydrates
As stated before, the crystallographic structure of gas hydrates at the Kedr MV is mainly due to the composition of thermogenic C2 in the volatile hydrocarbons. The concentration of C3, which is one of the sII-forming components, was two to three orders of magnitude smaller than that of C2, because biodegradation occurs and this preferentially reduces C3−5 of n-alkanes19,32, 33. The concentration of n-C4 was smaller than that of i-C4, whereas that of n-C5 was not detected (Table 1). C3 δ13C was around − 10‰, suggesting that light C3 is consumed by microbial activity. Assuming that sediment gas C3+ can be ignored, sediment gas ratio C1/C2 at the study area was 30 ± 17 (mean and standard deviation), and the concentration of C2 was ~ 3%. Such a composition of thermogenic gas is, therefore, considered to be supplied from a deep sediment layer, forming sI gas hydrates composed of mainly C1 and C211,12 in the lake floor sediment.
In the cases where sI gas hydrates plug and block migration pathways, upward fluid flow becomes more focused in other areas16. Once gas supply stops locally, gas hydrates begin to decompose, with the gas dissolving into gas–poor sediment pore water. In the system of C1 and C2, C2 is prone to be encaged in gas hydrate and decreases the equilibrium pressure of mixed-gas hydrate. Therefore, C2-rich gas hydrate forms in parallel with the decomposition of sI gas hydrate. The Colorado School of Mines Hydrate (CSMHYD) program34 showed that C2-rich sII gas hydrate (C2 concentration 17%) forms from mixed gas composed of C1 and C2 (C2 concentration 3%). The C2 concentration of hydrate-bound gas at the Kedr MV was ~ 14%, agreeing fairly well with the results of the CSMHYD program. Such secondary generation of gas hydrates can produce compositions and crystallographic structures that are different from the original crystals. A calorimetric study of synthetic C1 and C2 mixed-gas hydrate revealed that double peaks of heat flow correspond to the dissociation process of C1 and C2 mixed-gas hydrate, suggesting that C2-rich gas hydrate forms simultaneously from dissociated gas and showed that the second heat flow peak correspond to the dissociation of C2-rich gas hydrate18. The PXRD and solid-state 13C nuclear magnetic resonance techniques demonstrated that C2-rich sI gas hydrate forms in the dissociation process of C1 + C2 sII gas hydrate35.
Among twenty hydrate-bound cores in the Kedr area, four cores contained sI only, seven cores had sII only, and seven cores showed sII at the upper layer and sI at the lower layer, as observed at the Kukuy K-2 MV13,16,17. Furthermore, in the cores 2015St1GC15 and 2016St18GC2, gas hydrate structure had sI at the upper and lower layer, and sII at the middle layer. These results suggested that complex gas hydrate layers are composed of sI and sII in subsurface sediments as shown in the schematic illustration in Poort et al.16.
Depth profiles of C2 δ2H of gas hydrate cores from the Kedr MV are shown in Fig. 7. C2 δ2H of hydrate-bound gases varied between − 227‰ and − 206‰, with a grouping around − 210‰. C2 δ2H of sediment gases was also around − 210‰, indicating that C2 δ2H of the original thermogenic gas is − 210‰. As stated above, C2 δ2H of some cores showed low values at their base. Based on the isotopic fractionation of hydrogen in C2 during the formation of sI C2 hydrate36, δ2H of hydrate-bound C2 was 1.1‰ lower than that of residual C2. However, this is too small to explain the wide distribution in C2 δ2H shown in Fig. 7. On the other hand, Matsuda et al.37 reported that isotopic fractionation of hydrogen in C2 is dependent on the crystallographic structure: 1‰–2‰ for sI and ~ 10‰ for sII. Gas hydrates plotting around − 220‰ in C2 δ2H can be explained as a secondary generation of sII from dissociated gas hydrates, of which C2 δ2H was around − 210‰. However, some sII samples showed high C2 δ2H (around − 210‰), whereas some sI samples showed low C2 δ2H (around − 220‰). These results indicated that formation and dissociation processes of gas hydrates produce complicated isotopic profiles in C2 δ2H under non-equilibrium conditions.
Figure 7

Depth profiles of C2 δ2H of hydrate-bound and sediment gases. cmblf, centimetres below lake floor.

Full size image

Characteristics of hydrate-bound gases in sII
C3, i-C4, n-C4 and neo-C5 can be encaged in the larger hexadecahedral cages of sII1. n-C4 and neo-C5 can be encaged using a help gas (e.g. C1) to fill in the smaller dodecahedral cages of sII, because they cannot form pure n-C4 and neo-C5 hydrates, respectively. Figure 8 shows the concentration of C3, i-C4, n-C4, neo-C5 and i-C5 plotted against C2 concentration. The figure illustrates a clear division between sI (3–4%) and sII (14%) C2 concentrations. Data points between C2 concentrations of 5% and 13% were considered to have a mixture of sI and sII. Concentrations of C3, i-C4, n-C4 and neo-C5 had a positive correlation with the concentration of C2, and these concentrations in sII were 1 or 2 orders of magnitude larger than those in sI, suggesting that C3, i-C4, n-C4 and neo-C5 are encaged with C2 in the sII formation process.
Figure 8

Concentration of C3–5 against C2 concentration in the hydrate-bound gases.

Full size image

C3 values of 0.001%–0.01%, ~ 0.0001% of n-C4, and 0.0001%–0.01% of neo-C5 were also detected in sI hydrate-bound gas (Fig. 8), despite these hydrocarbons being unable to be encaged in sI. This can be explained by gases being adsorbed with sediments and gas hydrate crystals, which are then trapped in the grain boundary of polycrystalline gas hydrate crystals, and the gases are encaged if a small amount of sII crystals are present. For example, Uchida et al.38 examined natural gas hydrate retrieved at the Mackenzie Delta (onshore Canada) and detected C3 encaged in sII using Raman spectroscopy, although PXRD results suggested that the sample was sI and the major component of hydrate-bound gas was C1 (more than 99%).
neo-C5 is considered to form from the decomposition of gem-dimethylcycloalkanes derived from the terpenes of terrestrial organic matter39. It is easily enriched by preferential diffusion due to the nearly spherical molecules and its diffusion coefficient, which is higher than that of less branched isomers40. The sII hydrates retrieved at the Kukuy K-2 MV (central Baikal basin) contained 0.026–0.064% of neo-C5 in the volatile hydrocarbons13,14, and those at the Kedr MV had a maximum value of 0.054% of neo-C5 (Supplementary Information Table S1). On the contrary, in the case of natural gas hydrates retrieved at the Joetsu Basin (Japan Sea), neo-C5 was excluded and remained in sediment during the formation of sI gas hydrates from C1-rich gas41. The molecular size of i-C5 is considerably large to be encaged in the large cages of sII. Maximum concentration of i-C5 in the hydrate-bound gases was in several parts per million in both the fields of sI and sII (Fig. 8), indicating that i-C5 is not a hydrate-bound hydrocarbon and adsorbed with gas hydrate crystals and/or trapped in their grain boundary. More

General scheme
We reviewed the literature published until April 2018 on the economic impacts of invasive species. For reasons of feasibility (linguistic skills of the review team, restriction to a reasonable scale of the review), we conducted all searches in the English language assuming that a large body of knowledge (mostly from international peer-reviewed papers and reports) is written in English. The dates of each search process were systematically recorded. We used the following strategy for all repositories (Fig. 1), while also taking into consideration the specificity of their algorithms.
First, a literature search was performed using three online bibliographic sources successively to minimize the risk of omitting relevant materials (Fig. 1, step 1a): ISI Web of Science platform (https://webofknowledge.com/), Google Scholar database (https://scholar.google.com/) and the Google search engine (https://www.google.com/). We carefully composed appropriate search strings that were consensually retained as the most efficient among a set of potential candidates. A decision was taken following preliminary tests based on a handful of relevant articles provided by consulted subject experts on some taxonomic groups (amphibians, reptiles, fishes and ants). Final selection of search strings comprised those considered to have the largest potential to identify key references. Each search string was set to include a combination of two search terms, related to ‘invasive’ and ‘economics’. For both terms, we used a range of synonyms or related words. For example, for ‘invasive’ we used invasi*, invader or exotic; for ‘economics’, we used econom*, cost or monetary. In addition, the search string included exclusion terms to omit mismatches, for example, with studies from the field of medicine that are focused on pathologies or procedures that can be ‘invasive’ for patients. We complemented this search with documents gathered opportunistically (Fig. 1, step 1b). The potentially relevant materials derived from all these sources were combined in a single file and screened for duplicates. Second, retrieved documents were individually assessed at progressive levels (titles, then abstracts, keywords, and finally full text when abstracts were missing; Fig. 1, step 2) based on three criteria. Hence, materials were deemed relevant if (i) they matched with the linguistic competencies of the review team (i.e. written in English, or French where English language was restricted only to the title and/or abstract) for allowing reliable assessment, (ii) they contained at least one cost estimate (studies exclusively providing benefit estimates from direct use or exploitation of invasive species were excluded), and (iii) that this cost estimate is exclusively associated with invasive species (estimates merging non-invasive and invasive species, without the possibility of distinguishing the respective contribution of each group to the overall cost, were excluded). To ensure transparency and validity, each document was checked by two reviewers and in case of a disagreement between assessors, a third reviewer was involved. However, it was often difficult to judge from the topic whether the content of an article was relevant and so consequently many more articles were conservatively kept when final agreement was lacking among assessors.
Finally, relevant materials were scrutinized for data on economic costs (Table 1; Fig. 1, step 3). During this step, additional relevant materials were found as cited by the analysed materials. Obtained cost data were collated in a database and the costs were converted to a common and up to date currency (2017 US$), and then depicted by different descriptors. Categories extracted from relevant materials allow search of the database and data pre-selection to facilitating analysis of costs based on taxonomic groups, geographical areas, impacted sectors, types of costs, or other categories. The reliability of cost estimates and all associated information recorded in the final InvaCost database was systematically checked at least twice, and every ambiguous element was discussed to reach a consensus. We also checked all entries in the database to ensure that there were no obvious duplicate reports (i.e. multiple documents reporting the same cost estimate) or mistakes.
Hereafter, we specifically describe each of the steps made to generate InvaCost.
Literature search
Web of Science
We used the Web of Science (hereafter called WoS) to conduct a search for potentially relevant materials on 7 December 2017 (Fig. 1, step 1a). We applied the following search string: (econom* OR cost OR monetary OR dollar OR euro OR “sterling pound”) AND (invasi* OR alien OR non-indigenous OR nonindigenous OR nonnative OR non-native OR exotic OR introduced OR naturali* OR invader) NOT (cancer* OR cardio* OR surg* OR carcin* OR engineer* OR rotation OR ovar* OR polynom* OR purif* OR respirat* OR “invasive technique” OR carbon OR fuel OR therap* OR vehicle OR cell* OR drug OR fitness OR “operational research” OR banking OR liberalization). The terms were searched in the field code “Topic” which includes title, abstract and keywords, and which also comprises ‘Keywords Plus’ that are generated by WoS through an automatic computer algorithm, based on words and phrases that appear frequently in the titles of article’s bibliographic references and not necessarily in the main text of the article itself. To limit the search to relevant fields of research, we used the function ‘refine’ to exclude subject areas not related to economics and/or invasion biology.
We exported all records (n = 16,875) into an Excel worksheet30 (Table 1) to identify the relevant materials by a two-step procedure. First, we excluded the references identified only based on ‘Keywords Plus’, which were shown to be poor specific descriptors of the content of articles31. We also excluded references identified based on the presence of only a single search term in the topic, as we assumed that words related to both search terms (‘invasive’ and ‘economics’) should be mentioned at least once in the title, abstract and/or keywords of a relevant material. To identify these irrelevant materials within the references collected, we developed a script (see Code Availability) in the R programming language (R v.3.4.3)32. Subsequently, 10,592 references were kept for the next screening step based on the described criteria.
In the second step, the topic of every reference selected was checked manually to ensure potential relevance of its contents. This allowed the elimination of documents incorrectly identified as relevant, such as studies without a true monetary assessment, or those focusing on economic estimates not directly attributable to invasive species only. Finally, 1,333 documents were judged as relevant materials (Table 1) and moved to the final data collation step.
Google scholar
The Google Scholar database is a large source of grey as well as peer-reviewed literature. Nevertheless, we had to modify our approach in order to address inherent limitations of this database as a search tool (see Haddaway et al.33 for a comprehensive analysis). Typically, Google Scholar allows limited Boolean operators (no nesting using parentheses permitted) and search strings are limited to 256 characters. Additionally, only the first 1,000 search results can be viewed and the order in which results are returned is not disclosed. We also wanted to maximize novel information by avoiding too much overlap between the references collected with WoS and those gathered here.
In light of the above, we adapted our search string to generate the most efficient outcome, i.e. sufficiently pertinent to bring the most relevant items to the top of the result list while not unnecessarily large so as to limit the host of non-viewable results. Thus, the following search string was applied on 26 April 2018, using the advanced search facility to search for selected words anywhere in the article (see https://scholar.google.se/intl/en/scholar/help.html#searching for further details): dollars OR euros OR “USD” OR “EUR” OR “NZD” OR “AUD” OR “CAD” OR “GBP” OR “economic cost” OR “economic impact” OR “estimated cost” invasive species. We specified currencies for prioritising materials with monetary data in the top of the resulting list. These currencies were chosen as they were the most often used to express economic costs in the literature collected from the WoS. Nevertheless, any reference evoking economic costs in other currencies was expected to be also captured by some specific combinations of ‘economic’ terms in our search string that we would expect to be mentioned at least once in the full-text of relevant papers. In addition, we included the concomitant presence of ‘invasive’ and ‘species’ terms to restrict the outcomes to papers within the scope of our synthesis. Subsequently, we collected all viewable results (100 pages, n = 992 references of the 668,000 generated), thus going beyond the traditional and arbitrary sample size of first 50–100 results, which is frequently selected in many systematic reviews. We used a web-scraping programme (https://www.webscraper.io/) to extract all the titles’ references returned by the search in an Excel spreadsheet. Because we could not efficiently export the abstract for every reference, we screened them online to assess their potential relevance.
As a result of a search and relevance assessment within Google Scholar, the references, abstracts and specific bibliographic details of 432 documents were added to the sample for further analysis. After excluding duplicates with WoS retrieved references, 310 additional documents were included in the sample as potentially relevant materials (Table 1).
Google
We used the Google search engine to complete the standardised literature search. As when searching with Google Scholar, we took into account specific constraints related to the use of this search engine. Moreover, browsing through Google search results can be overwhelming due to the vast amount of information of highly variable quality. We attempted to implement a search strategy that could allow overcoming these limitations as much as possible. We used the following search string: economic species invasive OR nonnative OR alien OR exotic OR nonindigenous -disease -surgery -fungus -respiratory. We added four exclusion terms (disease, surgery, fungus, respiratory) identified during preliminary tests to restrict the number of irrelevant studies, associated with medical research. We did not use a range of economics-related terms, such as impact or cost, as they returned overly large numbers of mismatches.
The web search was conducted on 8 May 2018 by searching for specified terms within page titles of each document, in order to maximize the likelihood of identifying grey literature. We especially targeted grey literature because searches by the other two platforms mainly led to peer-reviewed publications. We assumed that documents published online by various governmental and non-governmental organisations (NGO), research centres and academic institutes are more likely to contain relevant data than other types of documents such as blogs and catalogues29. Therefore, we restricted our search to the documents located on governmental, academic and NGO webpages to ensure that explicit, traceable and expertise-based information was retrieved. We conducted independent searches for each type of webpage by specifying the type of web extension in the advanced search facility (.gov for governmental,.edu for academic, and.org for organisational webpages).
361 search hits were collected (document name, publishing year and URL of the main website homepage, if available) and stored in the database with the same host of dedicated information (Table 1). If the item analysed was a website homepage, we conducted on-line searches of potentially relevant materials within the website database(s), by filters if available, or by using the search bar with combinations of keywords. Websites that did not contain a database or search bar were searched manually. We then eliminated all duplicates resulting from references being listed on multiple websites, or due to typographical mistakes and/or incomplete records when reporting a reference within different repositories. A total of 119 potentially relevant materials was finally obtained (Table 1).
Targeted collection
Finally, we sourced other potentially relevant materials that did not originate from the above-described processes (Fig. 1, step 1b). On one side, we dedicated specific efforts on gathering cost estimates for particular taxa or areas for which data previously obtained seemed scarce. First, we made sure that some key species were adequately covered; for example, costs associated with invasive mosquito species responsible for much of the burden of mosquito-borne viral diseases worldwide (Aedes aegypti that mainly invaded the intertropical zone from the 15th-17th centuries, and Aedes albopictus for which the global dissemination was more recent34) were searched in a specific way using WoS and PubMed (https://www.ncbi.nlm.nih.gov/pubmed/) repositories (see supplementary file 1 for details on search strings and matching with PRISMA statements). Second, materials were also retrieved following requests to specialists (e.g. Aliens mailing list, https://list.auckland.ac.nz/sympa/info/aliens-l) to bridge gaps identified for Russia and China, two of the five largest countries for which available on-line data were particularly scarce. A typical message first summarized the objectives of our research project and second, requested recipients to provide relevant material and/or suggest further contacts in this regard. On the other side, we also compiled additional materials when establishing the methodology for the project (e.g. when testing different search string combinations at initial stages of the work), from the bibliographic alerts set up by the review team. All 1417 documents obtained from this process were entered in the database, with information on the person providing the document (Table 1;30). Subsequently, 150 documents identified as not previously retrieved were considered relevant for further, full-text screening (Table 1).
Extraction of cost estimates
The Online-only Table 1 comprises all the information of InvaCost that we mention further in this article, using simple quotation marks for ‘Columns’ of the database and italic letters for the different categories within each column. The full-text of each relevant material was scrutinized for any cost estimate that could be incorporated into InvaCost30. The final stage of inclusion/exclusion took place during this data extraction. When the screened documents reported cost estimates by citing sources that were not retrieved by our literature search, whenever possible we assessed the original sources of data in order to better characterize the reported cost. These novel information sources not initially captured by our literature search were then added to the collection list (Table 1). In such cases, we provided information on all documents that were consulted to trace back the original source (‘Previous materials’). In contrast, if no original cost data were found in the cited source, the document was discarded. For all reported costs where the original source was not available or accessible, we emphasized this in a dedicated column (‘Availability’).
Then, we first extracted raw cost data, i.e. how they appear in the material in local currency (‘Raw cost estimate local currency’). When multiple cost estimates were provided for a single instance, we calculated median values (e.g. different cost estimates according to several management scenarios dedicated to the same invasive population) and collated the minimum and maximum estimates provided (columns ‘Min/Max raw cost estimate local currency’). When costs were estimated at different time and/or spatial scales in the same material, we opted to choose – when possible – those estimate(s) that summarise(s) as effectively as possible the figure(s) shown in the study. If such an estimate was not obvious to identify throughout the full-text, we extracted every relevant cost estimate. In these latter cases where several cost estimates were provided in a single study, we also collated the minimum and maximum estimates provided.
Temporal information on the costs were also retrieved: the ‘Period of estimation’ as stated in the material and hence, when possible, the ‘Probable starting/ending year’ of the period of estimation and the ‘Time range’ (year if the estimate is given yearly or for a period up to one year, period if the estimate is given for a period exceeding a year). The ‘Occurrence’ column gives the status of the cost estimate as potentially ongoing (if the cost can be expected to continue beyond the period of estimation) or one-time (if the cost was deemed as unlikely to continue). For cost estimates provided without a clear indication on the timeframe considered, or covering periods shorter than a year, we considered them with a year ‘Time Range’ and a one-time ‘Occurrence’ to avoid the risk of overestimating the duration of collated costs. The ‘Raw cost estimate’– with complementary information on the ‘Time range’, ‘Period of estimation’ and ‘Occurrence’ – can be used to estimate total costs over a given period of time. We then transformed the raw cost estimates to cost estimates per year (‘Cost estimate per year’) by dividing the raw costs with a period ‘Time Range’ by the duration of the ‘Period of estimation’ (obtained from the difference between the ‘Probable ending year’ and ‘Probable starting year’). The raw costs with a year ‘Time Range’ were reported as they are, because they are already considered at the scale of a year.
Description of cost estimates in InvaCost
Each of the cost estimates recorded was characterized by a number of information, including (a) the reference from which the cost was extracted, (b) the taxonomy of the associated species, (c) the spatial and temporal coverage of the study, (d) the typology of each cost estimate and (e) the evaluation of the reliability of the estimation method(s). For most of the variables considered in InvaCost, a non-negligible part of the cost estimates was not attributable to a single existing category due to the lack of precise information provided by the authors or because they simultaneously belong to multiple categories. In such cases, we respectively reported them as either Diverse/Unspecified or as slash-separated lists of categories (e.g. Artiodactyla/Carnivora for the ‘Order’).
Details about the nature of the information retrieved as well as the choices made to characterize each cost are synthesized in Online-only Table 1:
(a) We provided bibliographic information on each reference (e.g. ‘Reference title’, ‘Authors’, ‘Publication year’). Others specific details (e.g. abstract, journal, download link) are given in a dedicated file30 with which the columns ‘Repository’ and ‘Reference ID’ of InvaCost allow correspondence of information.
(b) We normalised and harmonised all taxonomic information on the invasive species (‘Kingdom’ to ‘Species’ level) using the GBIF.org Backbone Taxonomy35. At this stage, spelling and other taxonomic errors were corrected. While each cost extracted was generally associated with a single invasive alien species, in some cases the data was related to multiple species without the possibility of disentangling species-specific costs. In this case, we mentioned either all species concerned if explicitly indicated by the author(s), or Diverse/Unspecified if not.
(c) We dedicated seven columns to describing the impacted area according to its environment (terrestrial and/or aquatic habitats), the temporal extent as mentioned earlier (e.g. ‘Period of estimation’, ‘Time range’) and the spatial coverage from the ‘Geographic region’ (e.g., Central America, South America, Oceania-Pacific Islands) – rather than the official continent for better accuracy – down to the exact site (‘Location’) when possible. Each area was related to its country of attachment, leading to some mismatches between the ‘Geographic region’ and ‘Official country’ columns due to the existence of countries with non-contiguous overseas territories. For instance, costs found from invaders in La Réunion (a French oversea department) were attributed to Africa as ‘Geographic region’ and France as ‘Country’, while France obviously belongs to European continent.
(d) We characterised the typology of each cost mainly based on the following descriptors. The ‘Implementation’ at the moment of the cost evaluation states whether the reported cost was observed (i.e. cost actually incurred by an invasive species within its invasive distribution area) or potential (i.e. not incurred but expected cost for an invasive species beyond its actual distribution area and/or predicted over time within or beyond its actual distribution area). The ‘Acquisition method’ provides information on how the cost data was obtained, i.e. report/estimation directly obtained or derived (using inference methods) from field-based information, or extrapolation relying on computational modelling. The ‘Impacted sectors’ indicates which activity, societal or market sectors were related to the cost estimate (see Table 2 for details). The ‘Type of cost’ ranges from the economic damages and losses incurred by an invasion (e.g. value of crop losses, damage repair) to different levels of means dedicated to the management of biological invaders (e.g. control, eradication, prevention).
Table 2 The different market and/or activity sectors mentioned in InvaCost.
Full size table

(e) Lastly, we evaluated the level of ‘Reliability’ of the methodology reported by the authors to provide cost estimates (Fig. 2). Prejudging the relevance of each cost estimate is not straightforward and could suffer from a high level of subjectivity. Here, we rather aimed to evaluate in the most objective manner whether the approach used for cost estimation was documented and traceable. Hence, materials that could not be accessed for full-text investigation were conservatively considered as of low reliability. Alternatively, each cost estimate recorded from any accessible material was qualitatively assessed as of high or low reliability following a procedure depending on the ‘Type of material’ analysed (peer-reviewed article or grey material; Fig. 2). Peer-reviewed articles and official documents (e.g. institutional or governmental reports) are likely validated by experts before publication. We assumed therefore that all cost estimates collected from these materials may likely be of high reliability. Conversely, for grey materials other than official reports, the attribution to one or other of these categories (high vs low reliability) was based on specific analysis of each cost estimate. We checked whether the method estimation was fully described, independently of its comprehensiveness, i.e. if the original sources or potential assumptions were properly documented or justified, and/or the calculation methodology was explicitly demonstrated. Here, we opted for a conservative strategy that might be not optimal, as depending mostly on the nature of the publication.
Fig. 2

Decision tree approach for assessing the reliability of the method used for estimating each cost. The colour of the boxes indicates which decision was taken: green when material was deemed as of high reliability, red when material was deemed as of low reliability, blue when taking any decision needs further investigation. The intended purpose of this process was not to evaluate the quality, relevance or realism of the studies performed for providing cost estimates, but rather to assess if the methodology (i) has been reviewed and validated by peers or experts prior any publication, or (ii) if not, whether this methodology was clearly stated and demonstrated.

Full size image

Beyond the factual elements included in the descriptors from (a) to (c), those presented in (d) and (e) (to which we can add the descriptor ‘Spatial scale’) are the result of a conceptual and analytical framework created based on our own experience. This experience was gained when collecting and getting acquainted with the diversity and complexity of situations one can find behind the “economic costs” linked to biological invasions, as well as the strategies used for estimating them. We think that the different subcategories identified therein (e.g. observed vs potential costs within the descriptor ‘Implementation’) should not be aggregated to limit potential confusions in future analysis. Also, we acknowledge that the possible sub-categories of these descriptors might be improved and adapted according to the scope of future analyses made using InvaCost. We are convinced that the descriptors thus defined and categorised may strongly help in this perspective.
Standardisation of cost data
Using definitions, data and indicators provided by the World Bank Open Data and the Organisation for Economic Cooperation and Development (OECD), we expressed all retrieved costs (raw costs and costs per year) in US dollars (US$) for the year 201730 using a multi-step procedure. We provided here two ways for standardising cost estimates according to the conversion factor: one based on the market exchange rate (local currency unit per US$, calculated as an annual average), and another based on the Purchasing Power Parity (PPP, local currency unit per US$, calculated as an annual average) that is the rate of currency conversion that standardises the purchasing power of different currencies by eliminating the differences in price levels between countries. Opting for one strategy or the other for further investigation or discussion is beyond the scope of this paper and will befall on the author(s) of future analyses made using InvaCost.
We first converted the cost estimates from local currencies to US$, by dividing the cost estimate with the official market exchange rate (https://data.worldbank.org/indicator/PA.NUS.FCRF?end=2017&start=1960) corresponding to the year of the cost estimation (‘Applicable year’, that is the year of the ‘Currency’ value, but not necessarily the year of the cost occurrence). The cost obtained in US$ of that year was then converted in 2017 US$ using an inflation factor that takes into account the evolution of the value of the US$ since the year of cost estimation. The inflation factor was computed by dividing the Consumer Price Index (CPI, which is a measure of the average change over time in the prices paid by consumers for a market basket of consumer goods and services; https://data.worldbank.org/indicator/FP.CPI.TOTL?end=2017&start=1960) of 2017 by the CPI of the year of the cost estimation.
As an alternative, we also converted costs to 2017 US$ value based on PPP instead of the classical market exchange rates in the initial conversion step. PPP values were primarily collected from data provided by the World Bank (https://data.worldbank.org/indicator/PA.NUS.PPP?end=2017&start=1990), or by the OECD (https://data.oecd.org/conversion/purchasing-power-parities-ppp.htm) when information was not retrievable through the World Bank database. For this purpose, we had to deal with published costs that were expressed in currency that was different from the country where the costs were estimated (e.g. published cost in African countries expressed in US or Canadian $). Thus, prior to using PPP as a conversion index, we had to perform a retro-conversion by multiplying the original cost estimate by the official market exchange rate (local currency unit per currency unit used). For PPP-based standardisation, it was not possible to perform the process for all cost estimates as PPP data do not exist for all countries and/or specific periods (we mentioned NA in the database when such information was missing).
In summary, we used the following formula to convert and standardise each cost estimate:

$${C}_{e}=left({{boldsymbol{M}}}_{{boldsymbol{V}}}/{{boldsymbol{C}}}_{{boldsymbol{F}}}right),times ,{{boldsymbol{I}}}_{{boldsymbol{F}}}$$

with Ce = Converted cost estimate (to 2017 US dollars based on exchange rate or Purchase Power Parity), MV = Cost estimate (either the ‘Raw cost estimate local currency’ extracted from analysed paper or the ‘Cost per year local currency’ transformed by us), CF = Conversion factor (either the official market exchange rate or the purchasing power parity, in US dollars), IF = Inflation factor since the year of cost estimation, calculated as CPI2017/CPIy with CPI corresponding to the Consumer Price Index and y corresponding to the year of the cost estimation (‘Applicable year’).
We thus provided four columns with the raw cost estimates or the cost estimates per year, expressed in 2017 USD based on the exchange rate or PPP.
Data summary
InvaCost currently contains 2419 cost estimates (1215 from peer-reviewed articles, 1204 from grey materials), collected from 849 references, of which 1769 estimates were deemed as of high reliability. In total, twenty currencies are reported in our database, the majority being US dollars, n = 1348 cost estimates. Not all cost estimates were successfully converted to 2017 US$ as (i) conversion data from official sources are available only since 1960 (cost estimates range from 1945 to 2017 in InvaCost) or simply not found for some years and countries, and/or (ii) cost data are sometimes simultaneously associated with several countries, constraining the PPP-based standardisations. Hence, respectively 2416 and 2126 estimates were successfully converted using market exchange rates and PPPs. Cost estimates are either direct reports/estimations (n = 2127) or values gathered from extrapolative computations (n = 292). At a taxonomic level, these estimates are associated with 343 species belonging to six kingdoms (Animalia, Bacteria, Chromista, Fungi, Plantae, Riboviria). InvaCost has global coverage (90 countries) and includes continental, insular and overseas territories. Data are associated with terrestrial as well as aquatic (freshwater, brackish and marine) environments. Costs were estimated at different spatial scales (continental (n = 35), country (n = 1111), global (n = 17), intercontinental (n = 9), regional (n = 67), site (n = 836), unit (n = 329)). The Table 3 summarises quantitative data and information reported in InvaCost for each geographic region considered (see also Supplementary file 2).
Table 3 Quantitative summary of information recorded in InvaCost according to the ‘Geographic region’ of the cost estimates.
Full size table

Possible applications
InvaCost is expected to help bridge the gap between a growing scientific understanding of invasion impacts and still inadequate management actions. This work is thus in line with the aims of a panel of decisions recently adopted by the Convention on Biological Diversity (Decision XIII/13, https://www.cbd.int/doc/decisions/cop-13/cop-13-dec-13-en.pdf) advocating the incorporation of invasion science knowledge into management planning. In addition to offer unique opportunities for future research, InvaCost will provide a strong quantitative and evidence-based support for impacts of invasive species reported in other databases such as the Global Register of Introduced and Invasive Species (GRIIS)20, helping refine information in this database. Also, invasive populations recorded in InvaCost but data deficient in the GRIIS should be ultimately classified in that database.
Additionally, InvaCost could be considered as another data-based component, adding novel and significant information on invader impacts categorised by the Socio-Economic Impact Classification of Alien Taxa (SEICAT)36. The latter is a classification system, applicable across a broad range of taxa and spatial scales, providing a consistent procedure for translating the broad range of measures and types of impacts into ranked levels of socio-economic impacts, assigning alien taxa on the basis of the best available evidence of their documented deleterious impacts. Quantitative support provided by InvaCost will strongly contribute to impact classification. Ultimately, integrating data from these diverse sources could allow a complete description of the overall impacts of biological invasions at regional and global scales.
Caveats and directions for further database improvement
Rather than claiming exhaustiveness of data collated, we highlight that InvaCost should be considered as the most current, standardised, accurate and globally representative repository of various economic losses and expenditures documented for the largest possible set of invaders. We are aware that our database can be improved in at least three ways.
First, InvaCost mostly does not include publications and reports not yet available in electronic format and/or using non-English language, leaving open the possibility of increasing data comprehensiveness and limiting potential biases. Indeed, local reports as well as research results from some countries (e.g., China, Russia) are likely to be published in non-English language37. Again, accessing grey literature is challenging as it is not systematically digitalised and/or included in well-curated bibliographic databases29. We strongly encourage future users of InvaCost to help gathering this currently unreachable information when possible. Furthermore, some mistakes might have occurred despite our best efforts when constructing InvaCost. In this regard, we advocate for regular public updates of InvaCost in order to improve it both quantitatively (by adding currently inaccessible or missed information) and qualitatively (if errors are identified).
Second, as the distribution and impacts of invaders are inherently dynamic for a number of reasons38, InvaCost should further consider the status of the species recorded for their economic impacts in order to improve both the relevance and the usefulness of the database. As an illustration, InvaCost likely includes invasive populations currently extirpated from particular areas after successful eradication campaign(s) as well as those still established but for which impacts are locally reduced as a result of management efforts. Attempting to obtain and integrate such information into InvaCost was beyond the scope of this work. Nonetheless, it should be reciprocally beneficial to establish connections between InvaCost and other databases such as the GRIIS that provides a harmonised, open source, multi-taxon database including verified information on the continued presence of introduced and invasive species for most countries20. In light of such additional information, the value of InvaCost will be its application for policy purposes, such as identification of exotic invaders that are currently associated with economic losses in particular areas. Also, crossing information between databases may allow the refinement of the descriptor ‘Spatial scale’ we propose here.
Third, we would recommend, for a future updated version of InvaCost that would require screening back all the materials, to improve the ‘Acquisition method’, ‘Implementation’ and ‘Reliability’ descriptors, to pay attention to the specificity of “avoided costs” and to create a new descriptor for ‘non-market values’. We detail these possibilities below.
Improving descriptors
An improved version of the ‘Acquisition method’ could lead to a subdivision of the extrapolation category into spatial, temporal and spatio-temporal extrapolation. This would allow simultaneous refinement from the currently binary ‘Implementation’ descriptor (observed vs potential) into several levels of certainty regarding the incurred cost (e.g. taking into consideration the temporality (past/current or predicted) of the onset of the cost and of the status of the invasive species in the study area). The next step for deeming the ‘Reliability’ of the cost estimates recorded in InvaCost would consist of assessing the repeatability of the methodology used, by adapting the approach previously developed by Bradshaw et al.14. The latter evidenced that assuming the reproducibility of published methods should not rely only on the nature of the materials and recognized the qualitative nature of the procedure, although applying this approach to InvaCost was constrained by the large sample size and high diversity in our database (Bradshaw et al.’s study focused on a single taxonomic class). Also, because InvaCost involves several collaborators and potential future contributors, consistent and objective criteria should be further defined to cope with the large array of materials, methods and situations encountered.
Avoided costs
Introducing certain actions against biological invasions leads to avoided costs. Such avoided costs are sometimes evaluated, for instance to examine the relevance of different potential actions or to assess the effectiveness of an action that was taken. However, avoided costs cover a great variety of situations and require a careful consideration for future analysis, even if they do not have to be analysed separately from the other economic costs gathered in InvaCost. For instance, in the case of hypothetical actions, avoided costs can be considered as minimum estimates of the “real” costs (if they are unknown). However, in the case of completed or planned actions, the reported data should be the original costs (if known) minus the avoided costs, because the latter do no longer exist. Some avoided costs are probably already included in InvaCost but they are likely underestimated because keywords such as “savings” or “benefits” were not included in the search strings. Also, even if they are sometimes mentioned as “benefits” in the literature, care should be taken not to confuse these avoided-costs with the benefits incurred by direct use or exploitation of invasive species. The latter have been ignored in InvaCost since they were relatively few (and beyond of the scope of this database), but might constitute a twin project.
A new ‘Non-market values’ descriptor
The means dedicated to preventing or managing an invasion (e.g. manual removal of invasive plants) and certain economic losses and damage due to an invasion (e.g. the value of crop losses or the repair costs of damaged infrastructures due to an invasive insect) are observable on markets. However, some costs are not observable on markets but can be translated in monetary terms using several valuation methods – for instance, the willingness to pay for the conservation of a native species that is impacted by an invasive species is considered as the value given by a group of people to preserving the native species (i.e. the value that would be lost if this native species was impacted). We recognize the importance of informing the public about “non-market values”, as giving an economic value to ecosystems or biodiversity can be a way of recognising and taking them into account in public decision-making processes39, but attention should be paid to the issues linked to their assessment40,41. Among others, the different methods for assessing non-market values do not necessarily capture the same aspects of the values, so the resulting estimates might be different. Moreover, the very principle of giving a value to “benefit from nature” through economic valuation is not necessarily acknowledged by the entirety of scientific and civil communities39,42. For future analysis, the ‘non-market values’ should not be systematically aggregated with the other economic costs gathered in InvaCost. It is to note that while some non-market values are probably already included in InvaCost within the losses and damage ‘Type of cost’, the loss of non-market values is probably largely underestimated in the database because they were not the primary focus of InvaCost and therefore the related keywords were not included in the search strings.
These possible ways of improvement call for completion and/or refinement of existing entries as well as integration of newly published or acquired data by future contributors in InvaCost, with the aim to consolidate its long-term relevance (cf. Usage Note paragraph). More

Tree-ring data
We used tree-ring records from over 210,000 trees of 110 species, distributed globally in habitats ranging from the tropics to the Arctic region over more than 70,000 sites (Supplementary Fig. 1, Supplementary Table 1). The largest publicly available data source from which we used data is the International Tree-Ring Data Bank (ITRDB, https://www.ncdc.noaa.gov/data-access/paleoclimatology-data/datasets/tree-ring). These were complemented with other datasets to maximize the number of records for each species and to fill in spatial gaps. A particularly large tree-ring dataset used in our analyses is the National Forestry Inventory data from the Ministère des Forêts de la Faune et des Parcs from Quebec, Canada52,53 (hereafter NFI-Quebec). This data consists of a complete set of ring-width data from 156,711 trees from 79.381 sites across the province of Quebec. Field tree-ring data were collected according to specific standard protocols52,53, which consisted of selection of up to nine trees in each plot, with 3–5 trees ( >91 mm diameter at breast height, DBH) randomly selected, 1–2 selected from the largest trees, and 1–2 from trees closest to the mean tree diameter of the plot52,53. From selected trees one core per tree was collected. Tropical tree-ring data were compiled from the ITRDB and from unpublished and published records26,27,54,55. For species with larger sample sizes, we distinguished between tree-ring data from trees that died naturally before the moment of sampling and trees that were alive at the moment of sampling, allowing us to test the assumption that living tree ages can be used to estimate trees natural lifespans (see section “Trade-off estimates and assessment of possible artefacts”). Part of these dead tree data were obtained from the ITRDB by selecting tree-ring records of which the last measured ring width was dated to before AD1900. We assumed that most of these trees must have been dead at the time of sampling as no records were collected before 1900. In addition, we compiled published dead tree data from refs. 21,56, and used subfossil tree-ring data from refs. 57,58. Supplementary Table 1 provides an overview of the datasets, and full details of each dataset are available online as supplemental info.
Various data controls and selection procedures were used to assure high confidence in our dataset. Where possible we tried to identify duplicate records, i.e., multiple cores taken from the same individual tree. This is only a problem for ITRDB, but not for the NFI-Quebec dataset where only one core per tree existed, or for datasets from co-authors. We thus merged ITRDB records that had identical ID′s except for the last character of their ID (e.g. 01a, 01b, or ID1-1, ID1-2, etc). From the ITRDB, we only used species for which we could obtain data from a minimum of 3 different sites with at least 20 records each, and only selected species that had a minimum total of 100 separate ring width series. We excluded those sites from the ITRDB that showed relative even age structures, and are thus unlikely to represent old-growth populations that provide robust estimates of trees’ maximum lifespans. To this end, we calculated for each ITRDB site the coefficient of variation in tree ages (CVAge = StandDevAge/ MeanAge × 100) and excluded sites with a CVAge lower than 10%. A large subset of ITRDB-data from 46 species has previously been inspected for data quality by co-author S. Voelker59. In this subset of data, each ring width series was manually re-aligned by cambial age (i.e., ring number from pith), providing more reliable estimates of tree ages. For the datasets that were not acquired from the ITRDB, we used slightly different criteria. From NFI-Quebec, we used all available sites, excluding those that were classified with evidence of recent management (commercial thinning or clear-cutting) and where fire or insect disturbances destroyed more than 25% of the forest cover.
For estimates of species-level early growth rates and lifespans (cf. Fig. 1a), we included only species with a minimum of 30 records, as lower sample size is unlikely to provide good approximations of tree life spans. This resulted in the inclusion of 110 species, with a median sample size of 305 trees and 12 sites per species. To assess within-species relationships between early growth and tree lifespan, we included 82 species. As a general rule, we included only species with more than a total of 150 trees and from at least 3 sites. About half of our species had more than 300 tree records (see Supplementary Information).
To assess what minimum sample size is needed to get a representative estimate of the true maximum age of a species or a site, and to evaluate how sample size affected estimates of trade-offs between early growth and longevity, we randomly resampled 500 times varying sample sizes—from 25 to 600 trees—from a subset of 11,752 Picea mariana trees from NFI-Quebec sites located north of 50.7°N. Comparison of the maximum ages of these random subsets of trees with the true observed maximum ages shows that a sample size of 100 trees results in 99.4% of the cases in maximum age estimates larger than the 95th percentile of the original dataset, and in 67.2% of the cases in ages larger than the 99th percentile original age (Supplementary Fig. 2a, b). As more than 70% of the species had at least 100 trees we thus assumed that for most species, the estimates of their lifespan were close to true lifespans. We used this same approach to assess how sample size affected the estimate of the trade-offs (i.e., estimation of the negative exponential decay constant; see next section for details). This analysis showed that sample size of 300 trees (corresponding to median sample sizes for trade-off analysis), leads to mean errors in the estimated slope of 12% (Supplementary Fig. 2c). Thus, for most species we achieve relatively accurate estimates of the trade-off strength. Low sample sizes for some species will nevertheless result in small errors of the mean slope, but we expect that positive and negative errors will cancel out against each other. Indeed, we do not observe a specific bias towards over- or under-estimation for low sample sizes, as the mean exponential decay constant for a simulated sample size of 150 trees is very similar to that observed (i.e., −0.409 versus −0.399).
Trade-off estimates and assessment of possible artefacts
The strength of the trade-offs between growth and tree lifespan was assessed for each species using a 95th quantile regression between mean early growth rate and the natural logarithm of age using the QUANTREG package in R60, as

$$begin{array}{*{20}{c}} {{{log}}(A_{(95{mathrm{th}},{mathrm{quantile}})}) = a + b cdot overline {{mathrm{RW}}} } \ {{mathrm{or}}} \ {{mathrm{Lifespan}} approx A_{left( {95{mathrm{th}},{mathrm{quantile}}} right)} = {mathrm{exp}},left( {a + b cdot overline {{mathrm{RW}}} } right)} end{array}$$
(1)

where A is age of the tree, (overline {{mathrm{RW}}}) is the mean ring width over the first 10 years. The constant b describes the negative exponential decay constant (i.e., exponential rate of decrease of tree lifespan with increasing early growth rate). This quantile regression fit results in similar estimates of the maximum ages of trees as the 95th percentile ages in binned early growth rate categories (see Supplementary Fig. 3a–c). Note that in contrast the maximum diameter does not vary strongly between slow and fast-growing trees (Supplementary Fig. 3d). We chose a relative short period, the first ten years, for estimating early growth as our study included some relative short-lived species. Previous studies have found similar results when using longer periods (50 years)56, and we expect no substantial difference using different early growth periods as tree growth is usually strongly auto-correlated in time61.
To assess trade-off strengths within species, we calculated the mean decay constant (b, Eq. 2) for each species using relative age, A/max(A), and relative mean early ring width, (overline {{mathrm{RW}}})/max((overline {{mathrm{RW}}})). Maximum of A and (overline {{mathrm{RW}}}) are species level maxima. The mean slope calculation across all species was weighted by the cube root of the sample size to account for the large differences between species in sample size, and confidence of the trade-off estimates.
While these relationships suggest true trade-offs, they may also be affected (or even driven by) the approaches or analytical methods used here. In particular, we here evaluate the effect of the following four possible artefacts on our results; (1) the use of living trees to estimate tree lifespans, (2) effect of recent growth increases on early growth-age relationship, (3) effects of pith offsets and wood decay on early growth-age relationship, (4) sampling artefacts, such as disproportionate selection of large trees.
(1)
Use of living trees: our analysis includes mostly trees that were sampled when still alive, and may thus not be representative for the true lifespan trees may achieve. To assess to which degree use of living trees may affect our results, we analyse and compare the trade-off strengths of trees that died before 1900 to living trees for 12 species with sufficient data availability (minimum of 150 dead and 150 living trees). As the slopes of dead and living trees do not differ significantly (Supplementary Fig. 5f, g, paired t-test exponential decay coefficient, t = −0.1095, p = 0.915, n = 12), we conclude that 95th quantile regressions on living trees can be used to approximate tree lifespan.

(2)
Effect of recent growth increases: recent growth stimulation of trees due e.g. to CO2 fertilisation, warming in higher latitudes, and/or nitrogen deposition, may result in observation of a trade-off. This is because recent increases in growth will lead to higher early growth rates for young trees compared to old trees, resulting in a negative relationship between early growth and tree age. The comparison of trade-off strength of dead versus living trees provides strong evidence that this effect does not drive the trade-off. In addition to this, we use a data driven forest simulation (see section “Examining the effects of growth stimulation on forest dynamics”) to assess how growth increases affect estimations of the trade-off strength. In this simulation, we used the actual tree-ring data to simulate realistic growth increases of Picea mariana tree-ring trajectories in response to high latitude warming. By sampling from these trajectories at the end of the growth increase period (i.e., year 350), and in a period without any recent growth increases (i.e., year 600, see Fig. 3e), we establish that growth increases result in only a small over-estimation of the trade-off, decreasing the exponential decay coefficient from −0.37 to −0.44 (see Supplementary Fig. 9c). Thus, it is unlikely that recent growth stimulation is the cause for the negative relation between early growth and tree lifespan.

(3)
Pith offset: tree-ring data, especially those acquired from ITRDB, may miss the innermost sections due to incomplete cores, decayed centres, or imperfect increment borer alignment. Missing rings will result in underestimation of tree ages and inaccurate early growth rates estimation and could thus affect the estimated relationship between early growth and lifespan. However, ring widths in most species decrease with tree age and size17, and even trees showing constant wood production with age, will show decreasing ring width because of geometry. Thus early growth in these samples will underestimate true growth rates and would most likely weaken the observed trade-off, rather than strengthening it. A comparison of species present in both the NFI-Quebec and the ITRDB datasets confirms this. The NFI-Quebec dataset was less affected by pith offset problems, as the trees were carefully screened and trees with substantial differences between cumulative ring widths and field diameters were excluded. Yet, we find that slopes were more negative for NFI-Quebec compared to ITRDB (mean b of −0.25 for Quebec vs. −0.10 for ITRDB, two-sided paired t.test, t = 2.49, p = 0.047, n = 7 species) and pith offsets thus do not explain the relationship. This comparison also shows that estimates of the strength of the trade-offs between early growth and longevity inferred from ITRDB data are probably conservative, as the Quebec data can be considered to be of higher quality, and were collected according to standard protocols. In contrast, data from the ITRDB may contain incomplete series and were collected for unknown purposes, and these issues probably weaken trade-offs in the ITRDB.

(4)
Sampling biases: one potential bias in our dataset may arise due to the tendency of tree-ring studies to sample predominantly large trees in the field (i.e., big tree selection bias62,63,64). This may result in a negative relationship between early growth and tree age, as young slow-growing trees tend to be underrepresented in the tree-ring sample (i.e., have not reached the field minimum size threshold yet), compared to fast-growing young trees that are much larger, and therefore more likely to be sampled. This effect would reduce the number of trees with slow early growth and young ages in the tree-ring sample (i.e., trees in the lower left-hand corner of the early growth-lifespan graphs, cf. Supplementary Fig. 6a), and results in overestimation of the 95th percentile age estimates for slow-growing trees. Our approach to estimate to which degree this bias affected our estimates of growth-lifespan trade-offs was as follows. We first used the tree-ring NFI-Quebec data of Picea mariana, combined with plot data from Quebec to reconstruct a new artificial tree-ring dataset with a size frequency distribution identical to the population size frequency distribution for this species in Quebec (Supplementary Fig. 6b). For each tree of Picea mariana sampled for their tree-rings we know the early growth rate and age, and also the complete diameter- and age-trajectory up to the year of sampling. From these data, we resampled for each size class (in bin widths of 2 cm) the same number of trees as that observed in the field. By doing this we filled in the lacking data of trees smaller than 91 mm, and created a new artificial tree-ring dataset that had an identical size structure to that observed in the field. We know the mean growth rate over the first ten years and the age at which each individual tree reached the diameter of their respective size class, and could thus reconstruct the early growth rate versus tree age graphs for the full population, including the smaller size classes which were missing from our original tree-ring sample. We then compared the early growth -lifespan relationship for the complete population to that of the trees larger than 91 mm, mimicking the NFI-Quebec field collection protocol. This shows that the exponential decrease is marginally larger (b = −0.505 compared to −0.470 for trees >91 mm) and that the intercept is lower (159 years compared to 220 for trees >91 mm) when sampling all trees compared to only trees with diameters >91 mm (Supplementary Fig. 6). Hence, the use of a minimum size threshold (91 mm) in the NFI from Quebec results in a slight underestimation of the trade-off (by ~7%) for the Picea mariana dataset. We also resampled from this artificial dataset the 10% largest trees, to mimic a hypothetical standard tree-ring sampling scenario that only samples the largest trees. Such a sampling scenario resulted in a decay constant of −0.432, thus again causing a small underestimation of the true trade-off. This simulation proves that the trade-off is not a result of a sampling bias.

Possible environmental drivers of the trade-offs
We evaluated whether the observed trade-off between early growth and tree lifespans could be caused by covariance of growth and lifespan with climate, soil or competition. Temperature variation for example reduces tree growth and lifespan in various species (cf. Fig. 2a, b). To this end, we calculated site-level mean early growth rates and the maximum tree age for a set of species covering different geographic regions (North America, Europe and Quebec). For Quebec, we combined multiple nearby locations to obtain a minimum of 30 trees per site, as sample sizes were low for each location. Site-level mean annual temperature and precipitation was obtained from WorldClim65. We then assessed for nine different species the effect of temperature and precipitation on site-level mean early life growth and maximum tree age using major axis regression from the package smart-366. These analyses confirm that early life growth is positively related to temperature for all nine species studied, and that lifespan decreases significantly with temperature for seven out of nine species (see Fig. 2a, b). Using linear mixed effect models with species as random factor (nlme-package-R67), we find that across all nine species, mean early life growth increases on average by 0.11 mm for each degree temperature increase, while lifespan decreases by 13 years for each degree temperature increase. Precipitation has no significant effect on early life growth or tree lifespan.
To disentangle whether lifespan decreases are a direct effect of temperature increases, or due to increases in early life growth, mixed effect models were run for all nine species that simultaneously included temperature and mean early life growth rate as explanatory variables for variation in tree lifespan. To account for species differences in growth and age, we used relative mean early ring width ((overline {RW})/max((overline {RW}))), and relative maximum age (A/max(A)), and used species as a random factor with random intercepts for both early life growth and temperature. This analysis shows that mean early life growth is a stronger predictor of tree lifespans than temperature (t-value early growth = −7.2, p  , D_0} end{array}} right..$$
(4)

Here b = 0.025, k = 3 × 10−11 and D0 = 91 mm, which is the minimum sampling diameter used for the NFI-Quebec. The rationale to set the mortality below D0 to zero in this model is that the trees sampled for tree rings did all survive to this diameter (as only trees with diameters >91 mm were cored). Thus, only by setting mortality to zero for trees , A_0} end{array}} right.,$$
(5)

with a = 0.021 and b = 0.0000015. Analogous to the diameter-dependent mortality model, we set mortality to zero for trees younger than 74 years (A0), which is the age at which the average Picea mariana tree reaches 91 mm in diameter (D0).
We next created a 600 year sequence of annually seeded, 1250 member, tree cohorts. Each member of a cohort is a randomly selected tree diameter growth trajectory derived from the tree-ring cores, extended in time to an age of 500 years. Short growth trajectories were extended by using the mean growth of the 10 oldest trees that had a similar ring width in the first ten years of growth. To this end all early mean ring width was grouped into six equal early ring width classes.
All trees of this so-constructed tree cohort set live, by construction, exactly 600 years. Realistic age structures were realised by sequentially (year-on-year and tree-by-tree) assigning death to trees where a random number generator identified those individuals  smaller than μ(D) or μ(A). To test the realism of this procedure, we compared the predicted and observed tree age versus early ring width relationship—or i.e. the growth-longevity trade-off. The slope of the relationship for the diameter-dependent mortality model μ(D) is very close to observed (Supplementary Fig. 9a), and is thus a realistic representation of the observed mortality process and justifies its use to examine the effect of a growth stimulation on standing stocks. In contrast, we find that the age-dependent mortality model μ(A) does not result in a significant trade-off between early growth and tree lifespan (Supplementary Fig. 9b), providing an ideal comparison for models that fail to incorporate the observed trade-offs.
To mimic growth stimulation, we boosted growth of trajectories from year t0 = 300 year onwards of the 600 year sequence of cohorts, while exposing the trajectories over the entire 600 year period to the mortality algorithm just described. We stimulated growth rate, (overline {{mathrm{RW}}}) (mm year−1) from year t0 = 300 year onwards according to

$$overline {{mathrm{RW}}_{stim}} left( t right) = overline {{mathrm{RW}}} left( t right) cdot exp left( {lambda left( A right) cdot delta Tleft( t right)} right),$$
(6)

and

$$delta Tleft( t right) = left{ {begin{array}{*{20}{c}} {frac{{overline {dT} }}{{dt}} cdot left( {t – t_0} right),} & {t – t_0 ,150 yrs) as younger trees are more sensitive to temperature increases than older trees (Supplementary Fig. 9d), and use an exponential model of the form

$$overline {{mathrm{RW}}} left( {A,,T} right) = a cdot {mathrm{exp}}left( {lambda left( A right) cdot delta T} right),$$
(8)

where a is a constant, λ(A) is the exponential increase rate for age class [A, A+ δA] and δT = (T − 0) (°C). We then used the exponential growth increase with temperature for each age band, λ(A), to estimate the relationship between tree age and λ(A) (Supplementary Fig. 9e). In all cases the age modulation of the stimulus is

$$lambda left( A right) = left{ {begin{array}{*{20}{c}} {0.0000132 cdot A^2-0.00291 cdot A + 0.229,} & {A , < ,135} \ {0.07,} & {A , ge ,135} end{array}} right..$$ (9) Finally, we compared the effect of growth stimulation on forests dynamics against simulation without growth increases (baseline) for the two different mortality models. We also performed one simulation where we multiplied the full growth series by 2, as a representation of the effect of growth stimulation on a faster-growing species. Finally, we evaluated for each simulation the mean ring width growth, the stem mortality rate, the age of the largest (75th percentile) trees that died and the change in the total basal area stock over time for the full population. For the stem mortality and the basal area stocks, we calculate and present the change in dynamics of the growth stimulation scenario relative to the baseline scenario without growth stimulation. All analyses and simulations were performed using R-studio, version 0.99.90370. Maps in SI figures were produced using the ggmap function from the R-package ‘ggplot2’71 More

Affiliations

State Key Laboratory for Biology of Plant Diseases and Insect Pests, Institute of Plant Protection, Chinese Academy of Agricultural Sciences, Beijing, China
Kris A. G. Wyckhuys & Yanhui Lu

Fujian Agriculture and Forestry University, Fuzhou, China
Kris A. G. Wyckhuys

University of Queensland, Brisbane, Queensland, Australia
Kris A. G. Wyckhuys & Michael J. Furlong

Chrysalis Consulting, Hanoi, Vietnam
Kris A. G. Wyckhuys

Zhejiang University, Hangzhou, China
Wenwu Zhou

CABI, Wallingford, UK
Matthew J. W. Cock & Frances E. Williams

USDA ARS, Maricopa, AZ, USA
Steven E. Naranjo

Secretariat of the Pacific Community SPC, Suva, Fiji
Atumurirava Fereti

Authors
Kris A. G. Wyckhuys

Yanhui Lu

Wenwu Zhou

Matthew J. W. Cock

Steven E. Naranjo

Atumurirava Fereti

Frances E. Williams

Michael J. Furlong

Corresponding authors
Correspondence to Kris A. G. Wyckhuys or Yanhui Lu or Wenwu Zhou. More

Multiple statistical methods have been developed to estimate the effect on birds and bats as a result of wind energy during the last 20 years26,27,28,29,30,31. Some of these studies are focused on the conservation status of the species32, the incidence factor of the wind turbines19,33, demographic parameters34,35, behavioural12 and also morphological parameters of the species36,37. In any case, it is essential to group all types of affections in order to be able to establish a global quantification that can be adapted to each species and to each specific wind farm. In other words, it can be obtained from a mathematical algorithm that allows quantifying the effect on each species, taking into account the characteristics of each wind farm3.
Furthermore, the formula that reflects the effect on the species must consider aspects related to the wind farm itself (type and distribution of turbines, occupation of the territory, etc.) and those related to the species, both in terms of its degree of threat and social importance, as well as its special sensitivity to the presence of the wind farm. According to this, the affection to each species must respond to the following formula:

$${text{AS}}_{{text{I}}} = {text{WF}}left( {{text{SS}}_{{text{I}}} + {text{VS}}_{{text{I}}} } right)$$

where ASI = Affection to species i, WF = Constant derived from the characteristics of the wind farm, SSI = Sensitivity of the species i to the presence of the wind farm, VSI = Social value of species i.
The index of affection, therefore, will be the product of multiplying the obtained values by the wind farm with those of each species that is present in the area.
Wind farm value constant (WF)
The impact value of the wind farm will be determined by the characteristics of the wind farm and also be influenced by both the characteristics of the wind farm (VWF) and its location (UF). At the same time, the VWF will be determined both by the affection of each wind turbine (WT) and by the distribution within the wind farm (extension and lines of turbines).

$${text{WF}} = {text{VWF}} + {text{UF}}$$

Value of the wind farm (VWF)
To calculate the overall effect of the wind farm not only is necessary to know the effect of each turbine but also its distribution in space. It is relevant to assess the distances between the wind turbine and if they are or not operating because when the turbines are very close together, the risk of moving between them is greater than in wind farms with more separate wind turbines38 and to know the number of rows in which the turbine are distributed. Crossing a wind farm with a single line of wind turbines is easier than those wind farms with several consecutive rows39. For this reason, the global affection of the wind farm (VWF) is understood as:
1.
The individual value of each of the wind turbine (WT) multiplied by the number of existing turbines.

2.
The total area occupied by the wind farm (AWF); in this way, it is not only considered the whole area of affection but also is established the density of the wind turbines.

3.
The number of rows that are included in the wind farm.

Based on the preceding information, the proposed formula for assessing the characteristics of the wind farm is:

$${text{VWF}} = left( {left( {{text{Ni}}*{text{WTi}}} right)/{text{AWF}}} right)^{{{1}/{text{F}}}}$$

where Ni: Number of wind turbines. WTi: Incidence of each wind turbines (the WT value will be the same for all unless in the same wind farm there were different types of wind turbines with different affection areas). AWF: Total surface of the area of study understood as the area formed by the vertical rectangle created between the furthest wind turbines from the same front line and the height of them. In the case of wind farms with more than one row, the total surface area is calculated as the sum of the surfaces of each row. F = Number of lines forming the wind farm.
Of these variables considered in the previous formula, it is only necessary to develop the affection inherent to each wind turbine (WT) that has to be calculated considering both the area of the turbine’s affection and the rotation speed of the blades.

$${text{WT}} = {text{AFM}}*{text{BRS}}$$

where AFM: Area of affection of each wind turbines, BRS: Blade rotation speed.
The area of affection of each turbine is the surface of the circumference formed by the blades (a), plus the surface of the triangle formed by the blades with the ground when they form an angle of 60° with the support tower (b), minus the intersection of both surfaces (c) (Fig. 3):

$${text{AFM}} = {text{a}} + {text{b}}{-}{text{c}}$$

Figure 3

Scheme and values to calculate the area of affection of each wind turbine. The area of affection of each turbine is the surface of the circumference formed by the blades (a), plus the surface of the triangle formed by the blades with the ground when they form an angle of 60° with the support tower (b), minus the intersection of both surfaces (c).

Full size image

Figure 4

Zoning scheme of risk areas. ZONE I: Corresponds to the free height between the ground and the blades. ZONE II: This zone corresponds to the area of the circumference formed by the blades when turning. ZONE III: Corresponds to the free height above the blades so that this interval is above the previous interval.

Full size image

Being: a = πr2   b = (sen60°*r)(L − cos 60° * r), where r is the length of the blades and L the height of the support tower. c = ((πr2/3) − ((sen60° * r)(cos 60° * r))).
To calculate the affection of the rotation speed of the blades (SB) it is assumed that the greater the rotation speed, the greater the turbulence and the greater the risk for the fauna19,40. In any case, this incidence is not linear but exponential since from a certain speed the affection can be considered high. To calculate this value, we established the following formula:

$${text{BRS}} = {1} + {text{Log}}left( {{text{SB}}} right)$$

Given that the value of the wind Farm (VWF) is the quotient between the sum of the areas affected by each turbine and the total area occupied, the value generally will be less than 1. In cases where the value is greater (when the surface area of the turbine multiplied by the rotation speed of the turbine is greater than the total surface of the area) it will be equal to 1. i.e., the maximum surface area affected cannot be greater than the surface area occupied by the total wind farm.
Location in the natural environment (UF)
Many works show the importance of selecting the location of the wind farm to minimize its impact on birdlife. However, it is possible that wind farms may be authorized in sensitive areas or in areas with poor environmental conditions (predominance of fog) or that have synergistic effects with adjacent wind farms. In this sense and as indicated in the introduction, there are four factors that can influence this impact: low visibility, proximity to sensitive areas, location in migratory crossings and synergies with other wind farms. Therefore, the value of this variable should be at least the same as that established for the previous variable (VWF). In this regard, it is proposed that the maximum value of the variables used to compute this factor should also be 1.

Visibility (VI): This variable measures the frequency of days with low visibility (fog, intense rain, etc.) compared with the total number of observation days (total number of days with low visibility/total number of observation days). The maximum value is 0.25.

Proximity to sensitive areas (ZS): Sensitive areas are those in which occur high concentrations of individuals, either because they are breeding areas, feeding areas, resting areas or roosts. Protected areas such as IBAS or LICs may also be considered. Not all species have the same radius of action, so setting a minimum radius of affection can only be established randomly. For example, for some species a radius of influence of 10 km is small but for others can be large. In any case and for having a uniform criterion, it will be considered that a sensitive area is close to the wind farm when it is located less than 10 km4, in this case, the value of this variable will be 0.25 and if they are between 10 and 50 km the value will be 0.15 while if it is more than 50 km is considered that the location of the wind farm does not influence these areas (value 0).

Migratory passes (MP): Migratory passes are those areas used by avian fauna for their daily or migratory movements. If the wind farm is located in one of these Migratory passes, the effect will be high so it will be valued with a maximum value (0.25) and the value will be minimal (0) if this is not the case.

Proximity to other wind farms (PWF): It is relevant to include this variable because of the proximity of different wind farms cause negative synergistic effects on the species by limiting the length of possible free corridors of wind turbines. In this way, the location of another wind farm less than 3 km away is considered very negative (0.25), between 3 and 5 km (0.15), between 5 and 10 km (0.10) and more than 10 km (0), it does not affect. If there is more than one wind farm in the area, the value will increase 0.25 if it is between 3 and 5 km and 0.15 if it is between 5 and 10 km.

$${text{UF}} = {text{VI}} + {text{ZS}} + {text{MP}} + {text{PWF}}$$

where WT: Value related to the location of the wind farm. VI: Predominant visibility in the area. ZS: Presence of sensitive areas in the vicinity of the wind farm. MS = Incidence of the wind farm in migratory crossings. PWF: Proximity to wind farms.

The possible maximum value for the wind farm location will be 1.
Therefore, the possible maximum value inherent in the characteristics and location of the wind farm will be 2. Substituting the values in the proposed formula:

$${text{WF}} = {text{VWF}} + {text{UF}}$$

And considering the values obtained for each mill, the wind farm in general and its location, the result is the following formula:

$${text{WF}} = left( {left( {{text{Ni}}*{text{IMi}}} right)/{text{AWF}}} right)^{{{1}/{text{F}}}} + left( {{text{VI}} + {text{ZS}} + {text{MP}} + {text{PWF}}} right)$$

Affection on the species
Not all species have the same sensitivity to the presence of the wind farm, being some of them more sensitive than others (25). On the other hand, the incidence on endangered species is not the same as that on species with stable populations in the area so, it is necessary to differentiate two types of variables related to the species: those related to the special sensitivity of each species to the presence of these infrastructures (SS) and the one inherent to its degree of threat, conservation or socioeconomic interest (VS). The affection value of each species will be the sum of the values of each type of variable. Therefore, the value of this section will be:

$${text{Affection}};{text{to}};{text{the}};{text{species}} = left( {{text{SS }} + {text{ VS}}} right)$$

Sensitivity of the species to the wind farm (SS)
These variables will be considered as the impact of the wind farm on each species due to its morphological, ethological, historical and demographic characteristics, etc. It is the closest thing to what could be understood as collision risk since it assesses the different characteristics of each bird (morphological, ethological, demographic, etc.) based on the risk of colliding with wind turbines and, valuing more those characteristics that enhance the probability of collision.
Bird size will be considered in this variable. A higher percentage of affection is detected on large birds in the majority of the recorder monitoring of the incidence of wind farms. However, this value seems to be overestimated since the detectability of carcasses of small birds and bats is lower as they remain less time on the ground30,41,42.
On the other hand, small birds show much less resistance to wind flows generated by the blades so it seems logical to think that the affection on this group of birds and on bats is higher. For this reason, a greater impact on small species has been assessed.
As a reference size, those birds smaller than or equal to a turtledove have been considered as small birds; medium-sized birds are those whose sizes are between a turtledove and a heron while those larger than a heron are considered as large birds. Considering these aspects:
The behaviour of different species will influence their risk of collision increasing the possibility of being affected by wind turbines38, for example, species that tend to go in groups show a greater risk of collision. The phenological characteristics of species are also important, for example, those species that are only in passage (prenuptial and postnuptial) will be little time in the study area but as they are not accustomed to the presence of wind turbines, probability of collision is high and possibly increased by going in large groups. Breeding species in the area are more dangerous because the young ones, still inexperienced in flight, show high risks of collision1. In other words, variables reflected in this section are related to the time the species spends in the area38, its dexterity in flight and its gregariousness. Together with these variables, the type of flight carried out by each species has been also considered: direct flights avoid staying longer in the area while cycloid or indirect movements increase the possibility of collision.

Seasonality: It considers the number of months in which the species is detected in the area. The maximum value is 1 if the species is sedentary (12 months) so each month is valued as 0.083.

Phenology: Marks the periods in which the species is present in the area. It is considered that species present in the breeding season or in passage show a greater risk than those that are only wintering. In this sense, if the species is in the breeding season will be valued with 0.75, only in winter 0.25 and 0.5 only in passage. When it appears in all seasons or in three of them, the value will be maximum (1). The value of the station will be also maximum if appears in two periods.

Flight height: In order to calculate flight height with risk for each species, the characteristics of each wind farm are considered. That is to say, they have to be adjusted to each wind farm since the interval of each zone will vary according to these ones. In this sense, for example, small birds that fly at lower altitudes can be located in the lower zone or not depending on the wind farm, just as large birds can be located in the area of the blades or above. In this sense, three zones have been established (Fig. 4):

ZONE I: Corresponds to the free height between the ground and the blades so, this interval goes from 0 m to the height resulting from subtracting the size of the blade from the length of the support tower. Value 0.5

ZONE II: This zone corresponds to the area with the greatest risk of collision since it is equivalent to the circumference formed by the blades when turning. Therefore, the interval will go from the previous height to its sum with the diameter of the circumference formed by the blades. Value: 1.

ZONE III: Corresponds to the free height above the blades so that this interval is above the previous interval. Value: 0.

When a species presents different flight heights, the one more frequent and that presents the greater risk will be selected.

Type of flight: Direct flights are considered to have a lower risk of collision than those that cause a longer stay in the area. The values will be 0.25 in direct flights and 0.5 in indirect flight.

Flock size: The risk of collision is considered higher when species show large groups so the following classification is established: One individual: 0.25; groups of 2–5 individuals: 0.5; groups of 6–10 individuals: 0.75; groups of more than 10 individuals: 1.

Historical variables (Maximum value 2).

A variable related to mortality detected in previous studies has also been included. Those species that are systematically detected in the mortality reviews of these infrastructures or exist high figures of mortality due to collision in specific wind farms should be considered.

Species with previous collision data (usual 2; medium 1; scarce 0.5; no record 0). This value is established at the discretion of the technician who performs the assessment, but as a habitual criterion, it can be considered as usual when the species appears in most of the studies (more than 30% of the studies), between 15 and 30% of the studies on average; and it will be classified as scarce if it only appears between 1 and 15% of the studies.

The last variables considered are related to the incidence on population parameters of each species. It has been considered that the species with reproductive strategy R suffer a lower incidence on the populations (although the mortality may be higher) since their reproductive efficiency partly solves this loss. However, species with K strategy suffer enormously when the mortality of young individuals is high. On the other hand, those species that frequently use the area where the wind turbines are located will show a higher probability of collision than those that are less common and those species that have high abundances in the area have also a higher probability of impact than those with few individuals16.

Survival-Fertility (type K or R) (K = 0.5; R = 0.2).

Frequency: This variable measures the frequency with which each species appears in the area in relation to the rest of the species present (total number of presences of the species/total number of presences detected). The maximum value is 1.

Abundance of the species in the area (number of individuals detected of the species i/total number of individuals detected of all species) (maximum value 1).

Species value (VS)
This value will include the conservation and socio-economic importance of the species (including the hunting value or social interest of some species). The affections on those species that are in a situation of greater risk of extinction must be considered in a relevant way, since the loss of a few individuals can represent the unfeasibility of the population. In this respect, both the degree of threat and the legal cataloguing of the different species have been considered.
The maximum value of this variable is much higher than the rest of variables since those species with the maximum protection value or degree of threat will have a value of 9. The cataloguing according to the Red Books will relate to the value established in Table 143. It has also been considered necessary to assess the socio-economic importance of some species. In this regard, it is taken into account not only the importance of hunting, which is relevant for some species of birds, but also its social importance, that is to say, those species which have conservation or recovery plans established in areas close to the different administrations or which are especially valued by the population, although their threat level is not very high (colonies of birds especially loved by the local population, etc.).
Table 1 Values given to the different classifications or threat level.
Full size table More

1.
Legras, J. L., Merdinoglu, D., Cornuet, J. M. & Karst, F. Bread, beer and wine: Saccharomyces cerevisiae diversity reflects human history. Mol. Ecol. 16, 2091–2102 (2007).
CAS  Article  Google Scholar 
2.
McGovern, P. E. et al. Fermented beverages of pre- and proto-historic China. Proc. Natl. Acad. Sci. U.S.A. 101, 17593–17598. https://doi.org/10.1073/pnas.0407921102 (2004).
ADS  CAS  Article  PubMed  PubMed Central  Google Scholar 

3.
Cavalieri, D., McGovern, P., Hartl, D., Mortimer, R. & Polsinelli, M. Evidence for S. cerevisiae fermentation in ancient wine. J. Mol. Evol. 57, S226–S232 (2003).
ADS  CAS  Article  Google Scholar 

4.
McGovern, P., Hartung, U., Badler, V., Glusker, D. & Exner, L. The beginnings of winemaking and viniculture in the ancient Near East and Egypt. Expedition 39, 3–21 (1997).
Google Scholar 

5.
Dudley, R. Ethanol, fruit ripening, and the historical origins of human alcoholism in primate frugivory. Integr. Comp. Biol. 44, 315–323 (2004).
CAS  Article  Google Scholar 

6.
Dudley, R. Fermenting fruit and the historical ecology of ethanol ingestion: is alcoholism in modern humans an evolutionary hangover?. Addiction 97, 381–388. https://doi.org/10.1046/j.1360-0443.2002.00002.x (2002).
Article  PubMed  Google Scholar 

7.
Carrigan, M. A. et al. Hominids adapted to metabolize ethanol long before human-directed fermentation. Proc. Natl. Acad. Sci. U.S.A. 112, 458–463. https://doi.org/10.1073/pnas.1404167111 (2015).
ADS  CAS  Article  PubMed  Google Scholar 

8.
Alba-Lois, L. & Segal-Kischinevzky, C. Yeast fermentation and the making of beer and wine https://www.nature.com/scitable/topicpage/yeast-fermentation-and-the-making-of-beer-14372813 (2010).

9.
Malacarne, M., Martuzzi, F., Summer, A. & Mariani, P. Protein and fat composition of mare’s milk: some nutritional remarks with reference to human and cow’s milk. Int. Dairy J. 12, 869–877. https://doi.org/10.1016/S0958-6946(02)00120-6 (2002).
CAS  Article  Google Scholar 

10.
Brady, M. First Taste. How Indigenous Australians Learned About Grog (Alcohol Education and Rehabilitation Foundation Ltd, Canberra, 2008).
Google Scholar 

11.
Brady, M. & McGrath, V. Making Tuba in the Torres Strait islands: the cultural diffusion and geographic mobility of an alcoholic drink. J. Pac. Hist. 45, 315–330. https://doi.org/10.1080/00223344.2010.530811 (2010).
Article  PubMed  Google Scholar 

12.
Varela, C. The impact of non-Saccharomyces yeasts in the production of alcoholic beverages. Appl. Microbiol. Biotechnol. 100, 9861–9874. https://doi.org/10.1007/s00253-016-7941-6 (2016).
CAS  Article  PubMed  Google Scholar 

13.
Jolly, N. P., Varela, C. & Pretorius, I. S. Not your ordinary yeast: non-Saccharomyces yeasts in wine production uncovered. FEMS Yeast Res. 14, 215–237. https://doi.org/10.1111/1567-1364.12111 (2014).
CAS  Article  PubMed  Google Scholar 

14.
Steinkraus, K. H. Handbook of Indigenous Fermented Foods, Second Edition, Revised and Expanded (Marcel Dekker, New York, 1995).
Google Scholar 

15.
Tamang, J. P., Watanabe, K. & Holzapfel, W. H. Review: diversity of microorganisms in global fermented foods and beverages. Front. Microbiol. https://doi.org/10.3389/fmicb.2016.00377 (2016).
Article  PubMed  PubMed Central  Google Scholar 

16.
Bahiru, B., Mehari, T. & Ashenafi, M. Yeast and lactic acid flora of tej, an indigenous Ethiopian honey wine: variations within and between production units. Food Microbiol. 23, 277–282. https://doi.org/10.1016/j.fm.2005.05.007 (2006).
CAS  Article  PubMed  Google Scholar 

17.
Vallejo, J. A. et al. Atypical yeasts identified as Saccharomyces cerevisiae by MALDI-TOF MS and gene sequencing are the main responsible of fermentation of chicha, a traditional beverage from Peru. Syst. Appl. Microbiol. 36, 560–564. https://doi.org/10.1016/j.syapm.2013.09.002 (2013).
CAS  Article  PubMed  Google Scholar 

18.
Puerari, C., Magalhães-Guedes, K. T. & Schwan, R. F. Physicochemical and microbiological characterization of chicha, a rice-based fermented beverage produced by Umutina Brazilian Amerindians. Food Microbiol. 46, 210–217. https://doi.org/10.1016/j.fm.2014.08.009 (2015).
CAS  Article  PubMed  Google Scholar 

19.
Escalante, A. et al. Characterization of bacterial diversity in Pulque, a traditional Mexican alcoholic fermented beverage, as determined by 16S rDNA analysis. FEMS Microbiol. Lett. 235, 273–279. https://doi.org/10.1016/j.femsle.2004.04.045 (2004).
CAS  Article  PubMed  Google Scholar 

20.
Lappe-Oliveras, P. et al. Yeasts associated with the production of Mexican alcoholic nondistilled and distilled Agave beverages. FEMS Yeast Res. 8, 1037–1052. https://doi.org/10.1111/j.1567-1364.2008.00430.x (2008).
CAS  Article  PubMed  Google Scholar 

21.
Jung, M. J., Nam, Y. D., Roh, S. W. & Bae, J. W. Unexpected convergence of fungal and bacterial communities during fermentation of traditional Korean alcoholic beverages inoculated with various natural starters. Food Microbiol. 30, 112–123. https://doi.org/10.1016/j.fm.2011.09.008 (2012).
Article  PubMed  Google Scholar 

22.
Greppi, A. et al. Determination of yeast diversity in ogi, mawe, gowe and tchoukoutou by using culture-dependent and -independent methods. Int. J. Food Microbiol. 165, 84–88. https://doi.org/10.1016/j.ijfoodmicro.2013.05.005 (2013).
CAS  Article  PubMed  Google Scholar 

23.
Spitaels, F. et al. The microbial diversity of traditional spontaneously fermented lambic beer. PLoS ONE https://doi.org/10.1371/journal.pone.0095384 (2014).
Article  PubMed  PubMed Central  Google Scholar 

24.
Tapsoba, F., Legras, J. L., Savadogo, A., Dequin, S. & Traore, A. S. Diversity of Saccharomyces cerevisiae strains isolated from Borassus akeassii palm wines from Burkina Faso in comparison to other African beverages. Int. J. Food Microbiol. 211, 128–133. https://doi.org/10.1016/j.ijfoodmicro.2015.07.010 (2015).
Article  PubMed  Google Scholar 

25.
Bokulich, N. A., Bamforth, C. W. & Mills, D. A. Brewhouse-resident microbiota are responsible for multi-stage fermentation of American coolship ale. PLoS ONE 7, e35507. https://doi.org/10.1371/journal.pone.0035507 (2012).
ADS  CAS  Article  PubMed  PubMed Central  Google Scholar 

26.
Bokulich, N. A., Thorngate, J. H., Richardson, P. M. & Mills, D. A. Microbial biogeography of wine grapes is conditioned by cultivar, vintage, and climate. Proc. Natl. Acad. Sci. U.S.A. 111, E139–E148. https://doi.org/10.1073/pnas.1317377110 (2014).
ADS  CAS  Article  PubMed  Google Scholar 

27.
Siren, K. et al. Taxonomic and functional characterization of the microbial community during spontaneous in vitro fermentation of Riesling must. Front. Microbiol. https://doi.org/10.3389/fmicb.2019.00697 (2019).
Article  PubMed  PubMed Central  Google Scholar 

28.
Morgan, H. H., du Toit, M. & Setati, M. E. The grapevine and wine microbiome: insights from high-throughput amplicon sequencing. Front. Microbiol. https://doi.org/10.3389/fmicb.2017.00820 (2017).
Article  PubMed  PubMed Central  Google Scholar 

29.
Williams, K. J. & Potts, B. M. The natural distribution of Eucalyptus species in Tasmania. Tasforests 8, 39–165 (1996).
Google Scholar 

30.
Calder, J. A. & Kirkpatrick, J. B. Climate change and other factors influencing the decline of the Tasmanian cider gum (Eucalyptus gunnii). Aust. J. Bot. 56, 684–692. https://doi.org/10.1071/BT08105 (2008).
Article  Google Scholar 

31.
Sanger, J. C., Davidson, N. J., O’Grady, A. P. & Close, D. C. Are the patterns of regeneration in the endangered Eucalyptus gunnii ssp. divaricata shifting in response to climate?. Austral. Ecol. 36, 612–620. https://doi.org/10.1111/j.1442-9993.2010.02194.x (2011).
Article  Google Scholar 

32.
Lahti, L. & Shetty, S. Tools for microbiome analysis in R. Version 1.9.1 https://microbiome.github.com/microbiome (2017).

33.
Morrison-Whittle, P. & Goddard, M. R. Quantifying the relative roles of selective and neutral processes in defining eukaryotic microbial communities. ISME J. 9, 2003–2011. https://doi.org/10.1038/ismej.2015.18 (2015).
CAS  Article  PubMed  PubMed Central  Google Scholar 

34.
Morrison-Whittle, P. & Goddard, M. R. From vineyard to winery: a source map of microbial diversity driving wine fermentation. Environ. Microbiol. 20, 75–84. https://doi.org/10.1111/1462-2920.13960 (2018).
Article  PubMed  Google Scholar 

35.
Brooker, M. I. H. A Key to Eucalypts in Britain and Ireland. (Forestry Commission Booklet 50: The Stationery Office, 1983).

36.
Forrest, M. & Moore, T. Eucalyptus gunnii: a possible source of bioenergy?. Biomass Bioenerg. 32, 978–980. https://doi.org/10.1016/j.biombioe.2008.01.010 (2008).
CAS  Article  Google Scholar 

37.
Guimarães, R. et al. Aromatic plants as a source of important phytochemicals: vitamins, sugars and fatty acids in Cistus ladanifer, Cupressus lusitanica and Eucalyptus gunnii leaves. Ind. Crop Prod. 30, 427–430. https://doi.org/10.1016/j.indcrop.2009.08.002 (2009).
CAS  Article  Google Scholar 

38.
Bugarin, D. et al. Essential oil of Eucalyptus gunnii hook. As a novel source of antioxidant, antimutagenic and antibacterial agents. Molecules 19, 19007–19020. https://doi.org/10.3390/molecules191119007 (2014).
CAS  Article  PubMed  PubMed Central  Google Scholar 

39.
Leborgne, N. et al. Introduction of specific carbohydrates into Eucalyptus gunnii cells increases their freezing tolerance. Eur. J. Biochem. 229, 710–717. https://doi.org/10.1111/j.1432-1033.1995.0710j.x (1995).
CAS  Article  PubMed  Google Scholar 

40.
Stuckel, J. G. & Low, N. H. The chemical composition of 80 pure maple syrup samples produced in North America. Food Res. Int. 29, 373–379. https://doi.org/10.1016/0963-9969(96)00000-2 (1996).
CAS  Article  Google Scholar 

41.
Taylor, M. W., Tsai, P., Anfang, N., Ross, H. A. & Goddard, M. R. Pyrosequencing reveals regional differences in fruit-associated fungal communities. Environ. Microbiol. 16, 2848–2858 (2014).
CAS  Article  Google Scholar 

42.
Pinto, C. et al. Wine fermentation microbiome: a landscape from different Portuguese wine appellations. Front. Microbiol. https://doi.org/10.3389/fmicb.2015.00905 (2015).
Article  PubMed  PubMed Central  Google Scholar 

43.
Miura, T., Sanchez, R., Castaneda, L. E., Godoy, K. & Barbosa, O. Is microbial terroir related to geographic distance between vineyards?. Environ. Microbiol. Rep. 9, 742–749. https://doi.org/10.1111/1758-2229.12589 (2017).
CAS  Article  PubMed  Google Scholar 

44.
Knight, S. J., Karon, O. & Goddard, M. R. Small scale fungal community differentiation in a vineyard system. Food Microbiol. https://doi.org/10.1016/j.fm.2019.103358 (2019).
Article  PubMed  Google Scholar 

45.
Tedersoo, L. et al. Global diversity and geography of soil fungi. Science 346, 1256688. https://doi.org/10.1126/science.1256688 (2014).
CAS  Article  PubMed  Google Scholar 

46.
Lin, Y. T., Whitman, W. B., Coleman, D. C. & Chiu, C. Y. Effects of reforestation on the structure and diversity of bacterial communities in subtropical low mountain forest soils. Front. Microbiol. https://doi.org/10.3389/fmicb.2018.01968 (2018).
Article  PubMed  PubMed Central  Google Scholar 

47.
Grangeteau, C. et al. Wine microbiology is driven by vineyard and winery anthropogenic factors. Microb. Biotechnol. 10, 354–370. https://doi.org/10.1111/1751-7915.12428 (2017).
CAS  Article  PubMed  Google Scholar 

48.
Burns, K. N. et al. Vineyard soil bacterial diversity and composition revealed by 16S rRNA genes: differentiation by geographic features. Soil Biol. Biochem. 91, 232–247. https://doi.org/10.1016/j.soilbio.2015.09.002 (2015).
CAS  Article  Google Scholar 

49.
Portillo, M. D. C., Franquès, J., Araque, I., Reguant, C. & Bordons, A. Bacterial diversity of Grenache and Carignan grape surface from different vineyards at Priorat wine region (Catalonia, Spain). Int. J. Food Microbiol. 219, 56–63. https://doi.org/10.1016/j.ijfoodmicro.2015.12.002 (2016).
Article  Google Scholar 

50.
Castaneda, L. E. & Barbosa, O. Metagenomic analysis exploring taxonomic and functional diversity of soil microbial communities in Chilean vineyards and surrounding native forests. PeerJ 5, e3098. https://doi.org/10.7717/peerj.3098 (2017).
CAS  Article  PubMed  PubMed Central  Google Scholar 

51.
Setati, M. E., Jacobson, D. & Bauer, F. F. Sequence-based analysis of the Vitis vinifera L. cv Cabernet Sauvignon grape must Mycobiome in three South African vineyards employing distinct agronomic systems. Front. Microbiol. https://doi.org/10.3389/fmicb.2015.01358 (2015).
Article  PubMed  PubMed Central  Google Scholar 

52.
Miura, T. et al. Shifts in the composition and potential functions of soil microbial communities responding to a no-tillage practice and bagasse mulching on a sugarcane plantation. Biol. Fertil. Soils 52, 307–322. https://doi.org/10.1007/s00374-015-1077-1 (2016).
CAS  Article  Google Scholar 

53.
Miura, T., Sanchez, R., Castaneda, L. E., Godoy, K. & Barbosa, O. Shared and unique features of bacterial communities in native forest and vineyard phyllosphere. Ecol. Evol. 9, 3295–3305. https://doi.org/10.1002/ece3.4949 (2019).
Article  PubMed  PubMed Central  Google Scholar 

54.
Hendgen, M. et al. Effects of different management regimes on microbial biodiversity in vineyard soils. Sci. Rep. https://doi.org/10.1038/s41598-018-27743-0 (2018).
Article  PubMed  PubMed Central  Google Scholar 

55.
Montecchia, M. S. et al. Pyrosequencing reveals changes in soil bacterial communities after conversion of Yungas forests to agriculture. PLoS ONE 10, 18. https://doi.org/10.1371/journal.pone.0119426 (2015).
CAS  Article  Google Scholar 

56.
Gleeson, D., Mathes, F., Farrell, M. & Leopold, M. Environmental drivers of soil microbial community structure and function at the Avon River Critical Zone Observatory. Sci. Total Environ. 571, 1407–1418. https://doi.org/10.1016/j.scitotenv.2016.05.185 (2016).
ADS  CAS  Article  PubMed  Google Scholar 

57.
Kemler, M. et al. Ion Torrent PGM as tool for fungal community analysis: a case study of Endophytes in Eucalyptus grandis reveals high taxonomic diversity. PLoS ONE https://doi.org/10.1371/journal.pone.0081718 (2013).
Article  PubMed  PubMed Central  Google Scholar 

58.
Piškur, J., Rozpędowska, E., Polakova, S., Merico, A. & Compagno, C. How did Saccharomyces evolve to become a good brewer?. Trends Genet. 22, 183–186. https://doi.org/10.1016/j.tig.2006.02.002 (2006).
CAS  Article  PubMed  Google Scholar 

59.
Lloyd, K. G., Steen, A. D., Ladau, J., Yin, J. Q. & Crosby, L. Phylogenetically novel uncultured microbial cells dominate Earth microbiomes. Msystems https://doi.org/10.1128/mSystems.00055-18 (2018).
Article  PubMed  PubMed Central  Google Scholar 

60.
Steen, A. D. et al. High proportions of bacteria and archaea across most biomes remain uncultured. ISME J. 13, 3126–3130. https://doi.org/10.1038/s41396-019-0484-y (2019).
Article  PubMed  PubMed Central  Google Scholar 

61.
Thrash, J. C. Culturing the uncultured: Risk versus reward. Msystems https://doi.org/10.1128/mSystems.00130-19 (2019).
Article  PubMed  PubMed Central  Google Scholar 

62.
Varela, C., Pizarro, F. & Agosin, E. Biomass content governs fermentation rate in nitrogen-deficient wine musts. Appl. Environ. Microbiol. 70, 3392–3400. https://doi.org/10.1128/Aem.70.6.3392-3400.2004 (2004).
CAS  Article  PubMed  PubMed Central  Google Scholar 

63.
Parker, M. et al. Factors contributing to interindividual variation in retronasal odor perception from aroma glycosides: The tole of odorant sensory detection threshold, oral microbiota, and hydrolysis in saliva. J. Agric. Food Chem. https://doi.org/10.1021/acs.jafc.9b05450 (2019).
Article  PubMed  Google Scholar 

64.
Bokulich, N. A. & Mills, D. A. Improved selection of internal transcribed spacer-specific primers enables quantitative, ultra-high-throughput profiling of fungal communities. Appl. Environ. Microbiol. 79, 2519–2526. https://doi.org/10.1128/AEM.03870-12 (2013).
CAS  Article  PubMed  PubMed Central  Google Scholar 

65.
Sternes, P. R., Lee, D., Kutyna, D. R. & Borneman, A. R. A combined meta-barcoding and shotgun metagenomic analysis of spontaneous wine fermentation. bioRxiv https://doi.org/10.1101/098061 (2017).
Article  Google Scholar 

66.
Bolger, A. M., Lohse, M. & Usadel, B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics 30, 2114–2120. https://doi.org/10.1093/bioinformatics/btu170 (2014).
CAS  Article  PubMed  PubMed Central  Google Scholar 

67.
Martin, M. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet J. https://doi.org/10.14806/ej.17.1.200 (2011).
Article  Google Scholar 

68.
Magoc, T. & Salzberg, S. L. FLASH: fast length adjustment of short reads to improve genome assemblies. Bioinformatics 27, 2957–2963. https://doi.org/10.1093/bioinformatics/btr507 (2011).
CAS  Article  PubMed  PubMed Central  Google Scholar 

69.
Mahe, F., Rognes, T., Quince, C., de Vargas, C. & Dunthorn, M. Swarm: robust and fast clustering method for amplicon-based studies. PeerJ 2, e593. https://doi.org/10.7717/peerj.593 (2014).
Article  PubMed  PubMed Central  Google Scholar 

70.
Caporaso, J. G. et al. QIIME allows analysis of high-throughput community sequencing data. Nat. Methods 7, 335–336. https://doi.org/10.1038/nmeth.f.303 (2010).
CAS  Article  PubMed  PubMed Central  Google Scholar 

71.
McMurdie, P. J. & Holmes, S. phyloseq: an R package for reproducible interactive analysis and graphics of microbiome census data. PLoS ONE 8, e61217. https://doi.org/10.1371/journal.pone.0061217 (2013).
ADS  CAS  Article  PubMed  PubMed Central  Google Scholar 

72.
Oksanen, J. et al. vegan: Community Ecology Package. R package version 2.5.4 https://CRAN.R-project.org/package=vegan (2019).

73.
Li, C., Yu, G. & Zhu, C. microbiomeViz—an R package for visualizing microbiome data https://github.com/lch14forever/microbiomeViz (2018).

74.
Kahle, D. & Wickham, H. ggmap: spatial visualization with ggplot2. R J. 5, 144–161 (2013).
Article  Google Scholar 

75.
Kassambara, A. ggpubr: ‘ggplot2’ based publication eady plots. R package version 0.2 https://CRAN.R-project.org/package=ggpubr (2018).

76.
Wickham, H. ggplot2: Elegant Graphics for Data Analysis (SpringerVerlag, New York, 2009).
Google Scholar 

77.
Team, R. C. R: a language and environment for statistical computing https://www.R-project.org/ (2017). More

1.
IPCC Special Report on Climate Change and Land (eds Shukla, P. R. et al.) (WMO and UNEP, 2019).
2.
Erb, K. H. et al. Unexpectedly large impact of forest management and grazing on global vegetation biomass. Nature 553, 73–76 (2018).
CAS  Article  Google Scholar 

3.
Searchinger, T. D., Wirsenius, S., Beringer, T. & Dumas, P. Assessing the efficiency of changes in land use for mitigating climate change. Nature 564, 249–253 (2018).
CAS  Article  Google Scholar 

4.
West, P. C. et al. Trading carbon for food: global comparison of carbon stocks vs. crop yields on agricultural land. Proc. Natl Acad. Sci. USA 107, 19645–19648 (2010).
CAS  Article  Google Scholar 

5.
Shepon, A., Eshel, G., Noor, E. & Milo, R. The opportunity cost of animal based diets exceeds all food losses. Proc. Natl Acad. Sci. USA https://doi.org/10.1073/pnas.1713820115 (2018).

6.
Poore, J. & Nemecek, T. Reducing food’s environmental impacts through producers and consumers. Science 992, 987–992 (2018).
Article  Google Scholar 

7.
Tilman, D. & Clark, M. Global diets link environmental sustainability and human health. Nature 515, 518–522 (2014).
CAS  Article  Google Scholar 

8.
Springmann, M. et al. Health and nutritional aspects of sustainable diet strategies and their association with environmental impacts: a global modelling analysis with country-level detail. Lancet Planet. Health 2, e451–e461 (2018).
Article  Google Scholar 

9.
Herrero, M. et al. Greenhouse gas mitigation potentials in the livestock sector. Nat. Clim. Change 6, 452–461 (2016).
Article  Google Scholar 

10.
Batchelor, J. L., Ripple, W. J., Wilson, T. M. & Painter, L. E. Restoration of riparian areas following the removal of cattle in the northwestern great basin. Environ. Manage. 55, 930–942 (2014).
Article  Google Scholar 

11.
Sitters, J., Kimuyu, D. M., Young, T. P., Claeys, P. & Olde Venterink, H. Negative effects of cattle on soil carbon and nutrient pools reversed by megaherbivores. Nat. Sustain. 3, 360–366 (2020).
Article  Google Scholar 

12.
Alexandratos, N. & Bruinsma, J. World Agriculture Towards 2030/2050: The 2012 Revision (FAO, 2012).

13.
Willett, W. et al. Food in the Anthropocene: the EAT–Lancet Commission on healthy diets from sustainable food systems. Lancet 6736, 3–49 (2019).
Google Scholar 

14.
IPCC Special Report on Global Warming of 1.5 °C (eds Masson-Delmotte, V. et al.) (WMO, 2018).

15.
Fry, J. P., Mailloux, N. A., Love, D. C., Milli, M. C. & Cao, L. Feed conversion efficiency in aquaculture: do we measure it correctly? Environ. Res. Lett. 13, 024017 (2018).
Article  Google Scholar 

16.
Van Zanten, H. H. E. et al. Defining a land boundary for sustainable livestock consumption. Glob. Change Biol. https://doi.org/10.1111/gcb.14321 (2018).

17.
Griscom, B. W. et al. Natural climate solutions. Proc. Natl Acad. Sci. USA 114, 11645–11650 (2017).
CAS  Article  Google Scholar 

18.
Randerson, J. T. et al. Multicentury changes in ocean and land contributions to the climate–carbon feedback. Glob. Biogeochem. Cycles 29, 744–759 (2015).
CAS  Article  Google Scholar 

19.
Smith, P. et al. How much land-based greenhouse gas mitigation can be achieved without compromising food security and environmental goals? Glob. Change Biol. 19, 2285–2302 (2013).
Article  Google Scholar 

20.
Schmidinger, K. & Stehfest, E. Including CO2 implications of land occupation in LCAs-method and example for livestock products. Int. J. Life Cycle Assess. 17, 962–972 (2012).
CAS  Article  Google Scholar 

21.
Stehfest, E. et al. Climate benefits of changing diet. Clim. Change 95, 83–102 (2009).
CAS  Article  Google Scholar 

22.
Ramankutty, N., Evan, A. T., Monfreda, C. & Foley, J. A. Farming the planet: 1. Geographic distribution of global agricultural lands in the year 2000. Glob. Biogeochem. Cycles 22, GB1003 (2008).
Article  Google Scholar 

23.
Monfreda, C., Ramankutty, N. & Foley, J. A. Farming the planet: 2. Geographic distribution of crop areas, yields, physiological types, and net primary production in the year 2000. Glob. Biogeochem. Cycles 22, GB1022 (2008).
Article  Google Scholar 

24.
Cassidy, E. S., West, P. C., Gerber, J. S. & Foley, J. A. Redefining agricultural yields: from tonnes to people nourished per hectare. Environ. Res. Lett. 8, 034015 (2013).
Article  Google Scholar 

25.
Bouwman, A. F., Van der Hoek, K. W., Eickhout, B. & Soenario, I. Exploring changes in world ruminant production systems. Agric. Syst. 84, 121–153 (2005).
Article  Google Scholar 

26.
Hansen, M. C. et al. High-resolution global maps of 21st-century forest cover change. Science 342, 850–853 (2013).
CAS  Article  Google Scholar 

27.
Herrero, M. et al. Biomass use, production, feed efficiencies, and greenhouse gas emissions from global livestock systems. Proc. Natl Acad. Sci. USA 110, 20888–20893 (2013).
CAS  Article  Google Scholar 

28.
Erb, K. H. et al. Biomass turnover time in terrestrial ecosystems halved by land use. Nat. Geosci. 9, 674–678 (2016).
CAS  Article  Google Scholar 

29.
Fetzel, T. et al. Quantification of uncertainties in global grazing systems assessment. Glob. Biogeochem. Cycles 31, 1089–1102 (2017).
CAS  Article  Google Scholar  More

1.
Plastics Europe. Plastics—the Facts 2019; 2019. https://www.plasticseurope.org/application/files/9715/7129/9584/FINAL_web_version_Plastics_the_facts2019_14102019.pdf.
2.
Ritchie H, Roser M. Plastic pollution. Our World in Data; 2018. https://ourworldindata.org/plastic-pollution.

3.
UNEP. Single-use plastics: a roadmap for sustainability; 2018. https://wedocs.unep.org/bitstream/handle/20.500.11822/25496/singleUsePlastic_sustainability.pdf?sequence=1&isAllowed=y.

4.
Geyer R, Jambeck JR, Law KL. Production, use, and fate of all plastics ever made. Sci Adv. 2017;3:e1700782. https://doi.org/10.1126/sciadv.1700782.
CAS  Article  PubMed  PubMed Central  Google Scholar 

5.
Jambeck JR, Geyer R, Wilcox C, Siegler TR, Perryman M, Andrady A, et al. Plastic waste inputs from land into the ocean. Science. 2015;347:768–71. https://doi.org/10.1126/science.1260352.
CAS  Article  PubMed  Google Scholar 

6.
Derraik JGB. The pollution of the marine environment by plastic debris: a review. Mar Pollut Bull. 2002;44:842–52.
CAS  PubMed  Google Scholar 

7.
World Economic Forum. Ellen MacArthur Foundation. The new plastics economy: rethinking the future of plastics; 2016. www3.weforum.org/docs/WEF_The_New_Plastics_Economy.pdf.

8.
Stelfox M, Hudgins J, Sweet M. A review of ghost gear entanglement amongst marine mammals, reptiles and elasmobranchs. Mar Pollut Bull. 2016;11:6–17.
Google Scholar 

9.
Rios LM, Moore C, Jones PR. Persistent organic pollutants carried by synthetic polymers in the ocean environment. Mar Pollut Bull. 2007;54:1230–7.
CAS  PubMed  Google Scholar 

10.
Rochman CM, Hoh E, Kurobe T, Teh SJ. Ingested plastic transfers hazardous chemicals to fish and induces hepatic stress. Sci Rep. 2013;3:32632. https://doi.org/10.1038/srep03263.
Article  Google Scholar 

11.
Setala O, Fleming-Lehtinen V, Lehtiniemi M. Ingestion and transfer of microplastics in the planktonic food web. Environ Pollut. 2014;185:77–83.
CAS  PubMed  Google Scholar 

12.
Lucas N, Bienaime C, Belloy C, Queneudec M, Silvestre F, Nava-Saucedo JE. Polymer biodegradation: mechanisms and estimation techniques. Chemosphere. 2008;73:429–42.
CAS  PubMed  Google Scholar 

13.
Harrison JP, Boardman C, O’Callaghan K, Delort AM, Song J. Biodegradability standards for carrier bags and plastic films in aquatic environments: a critical review. R Soc Open Sci. 2018;5:171792. https://doi.org/10.1098/rsos.171792.
CAS  Article  PubMed  PubMed Central  Google Scholar 

14.
Sudesh K, Abe H, Doi Y. Synthesis, structure and properties of polyhydroxyalkanoates: biological polyesters. Prog Polym Sci. 2000;25:1503–55.
CAS  Google Scholar 

15.
Labet M, Thielemans W. Synthesis of polycaprolactone: a review. Chem Soc Rev. 2009;38:3484–504.
CAS  PubMed  Google Scholar 

16.
Tserki V, Matzinos P, Pavlidou E, Vachliotis D, Panayiotou C. Biodegradable aliphatic polyesters. Part I. Properties and biodegradation of poly(butylene succinate-co-butylene adipate). Polym Degrad Stabil. 2006;91:367–76.
CAS  Google Scholar 

17.
Yamamoto M, Witt U, Skupin G, Beimborn D, Müller J. Biodegradable aliphatic-aromatic polyesters: “Ecoflex®”. Biopolymers Online; 2005.

18.
Nampoothiri KM, Nair NR, John RP. An overview of the recent developments in polylactide (PLA) research. Bioresour Technol. 2010;101:8493–501.
Google Scholar 

19.
Ku H, Wang H, Pattarachaiyakoop N, Trada M. A review on the tensile properties of natural fiber reinforced polymer composites. Compos B Eng. 2011;42:856–73.
Google Scholar 

20.
Jendrossek D, Schirmer A, Schlegel HG. Biodegradation polyhydroxyalkanoic acids. Appl Microbiol Biot. 1996;46:451–63.
CAS  Google Scholar 

21.
Kasuya K, Ohura T, Masuda K, Doi Y. Substrate and binding specificities of bacterial polyhydroxybutyrate depolymerases. Int J Biol Macromol. 1999;24:329–36.
CAS  PubMed  Google Scholar 

22.
Tokiwa Y, Suzuki T. Hydrolysis of polyesters by lipases. Nature. 1977;270:76–78.
CAS  PubMed  Google Scholar 

23.
Jaeger KE, Steinbuchel A, Jendrossek D. Substrate specificities of bacterial polyhydroxyalkanoate depolymerases and lipases: bacterial lipases hydrolyze poly(omega-hydroxyalkanoates). Appl Environ Microbiol. 1995;61:3113–8.
CAS  PubMed  PubMed Central  Google Scholar 

24.
Murphy CA, Cameron JA, Huang SJ, Vinopal RT. Fusarium polycaprolactone depolymerase is cutinase. Appl Environ Microbiol. 1996;62:456–60.
CAS  PubMed  PubMed Central  Google Scholar 

25.
Maeda H, Yamagata Y, Abe K, Hasegawa F, Machida M, Ishioka R, et al. Purification and characterization of a biodegradable plastic-degrading enzyme from Aspergillus oryzae. Appl Microbiol Biotechnol. 2005;67:778–88.
CAS  PubMed  Google Scholar 

26.
Murphy CA, Cameron JA, Huang SJ, Vinopal RT. A second polycaprolactone depolymerase from Fusarium, a lipase distinct from cutinase. Appl Microbiol Biotechnol. 1998;50:692–6.
CAS  Google Scholar 

27.
Ando Y, Yoshikawa K, Yoshikawa T, Nishioka M, Ishioka R, Yakabe Y. Biodegradability of poly(tetramethylene succinate-co-tetramethylene adipate): I. Enzymatic hydrolysis. Polym Degrad Stabil. 1998;61:129–37.
CAS  Google Scholar 

28.
Shinozaki Y, Morita T, Cao X, Yoshida S, Koitabashi M, Watanabe T, et al. Biodegradable plastic-degrading enzyme from Pseudozyma antarctica: cloning, sequencing, and characterization. Appl Microbiol Biotechnol. 2013;97:2951–9.
CAS  PubMed  Google Scholar 

29.
Muroi F, Tachibana Y, Soulenthone P, Yamamoto K, Mizuno T, Sakurai T, et al. Characterization of a poly(butylene adipate-co-terephthalate) hydrolase from the aerobic mesophilic bacterium Bacillus pumilus. Polym Degrad Stabil. 2017;137:11–22.
CAS  Google Scholar 

30.
Tokiwa Y, Calabia BP. Biodegradability and biodegradation of poly(lactide). Appl Microbiol Biotechnol. 2006;72:244–51.
CAS  PubMed  Google Scholar 

31.
Williams DF. Enzymic hydrolysis of polylactic acid. Eng Med. 1981;10:5–7.
Google Scholar 

32.
Nakayama A, Yamano N, Kawasaki N. Biodegradation in seawater of aliphatic polyesters. Polym Degrad Stabil. 2019;166:290–9.
CAS  Google Scholar 

33.
Luzier WD. Materials derived from biomass biodegradable materials. Proc Natl Acad Sci USA. 1992;89:839–42.
CAS  PubMed  Google Scholar 

34.
Kasuya K, Takagi K, Ishiwatari S, Yoshida Y, Doi Y. Biodegradabilities of various aliphatic polyesters in natural waters. Polym Degrad Stabil. 1998;59:327–32.
CAS  Google Scholar 

35.
Teramoto N, Urata K, Ozawa K, Shibata M. Biodegradation of aliphatic polyester composites reinforced by abaca fiber. Polym Degrad Stabil. 2004;86:401–9.
CAS  Google Scholar 

36.
Sekiguchi T, Saika A, Nomura K, Watanabe T, Fujimoto Y, Enoki M, et al. Biodegradation of aliphatic polyesters soaked in deep seawaters and isolation of poly(ε-caprolactone)-degrading bacteria. Polym Degrad Stabil. 2011;96:1397–403.
CAS  Google Scholar 

37.
Fujimaki T. Processability and properties of aliphatic polyesters, ‘BIONOLLE’, synthesized by polycondensation reaction. Polym Degrad Stabil. 1998;59:209–14.
CAS  Google Scholar 

38.
Witt U, Müller RJ, Deckwer WD. New biodegradable polyester-copolymers from commodity chemicals with favorable use properties. J Environ Polym Degrad. 1995;3:215–23.
CAS  Google Scholar 

39.
Bagheri AR, Laforsch C, Greiner A, Agarwal S. Fate of so‐called biodegradable polymers in seawater and freshwater. Glob Chall. 2017;1:1700048. https://doi.org/10.1002/gch2.201700048.
Article  PubMed  PubMed Central  Google Scholar 

40.
Zambrano MC, Pawlak JJ, Daystar J, Ankeny M, Goller CC, Venditti RA. Aerobic biodegradation in freshwater and marine environments of textile microfibers generated in clothes laundering: effects of cellulose and for polyester-based microfibers on the microbiome. Mar Pollut Bull. 2020;151:110826. https://doi.org/10.1016/j.marpolbul.2019.110826.
CAS  Article  PubMed  Google Scholar 

41.
Taylor LE, Henrissat B, Coutinho PM, Ekborg NA, Hutcheson SW, Weiner RA. Complete cellulase system in the marine bacterium Saccharophagus degradans strain 2-40T. J Bacteriol. 2006;188:3849–61.
CAS  PubMed  PubMed Central  Google Scholar 

42.
Yang JC, Madupu R, Durkin SA, Ekborg NA, Pedamallu CS, Hostetler JB, et al. The Complete genome of Teredinibacter turnerae T7901: an intracellular endosymbiont of marine wood-boring bivalves (shipworms). PLoS ONE. 2009;4:e6085. https://doi.org/10.1371/journal.pone.0006085.
CAS  Article  PubMed  PubMed Central  Google Scholar 

43.
Sawant SS, Salunke BK, Taylor LE, Kim BS. Enhanced agarose and xylan degradation for production of polyhydroxyalkanoates by co-culture of marine bacterium, Saccharophagus degradans and its contaminant, Bacillus cereus. Appl Sci. 2017;7. https://doi.org/10.3390/app7030225.

44.
Liu G, Wu SM, Jin WH, Sun CM. Amy63, a novel type of marine bacterial multifunctional enzyme possessing amylase, agarase and carrageenase activities. Sci Rep. 2016;6:18726. https://doi.org/10.1038/srep18726.
CAS  Article  PubMed  PubMed Central  Google Scholar 

45.
Ramesh S, Mathivanan N. Screening of marine actinomycetes isolated from the Bay of Bengal, India for antimicrobial activity and industrial enzymes. World J Microb Biotechnol. 2009;25:2103–11.
CAS  Google Scholar 

46.
Nogi Y, Yoshizumi M, Miyazaki M. Thalassospira povalilytica sp. nov., a polyvinyl-alcohol-degrading marine bacterium. Int J Syst Evol Microbiol. 2014;64:1149–53.
CAS  PubMed  Google Scholar 

47.
Sedlacek P, Slaninova E, Enev V, Koller M, Nebesarova J, Marova I, et al. What keeps polyhydroxyalkanoates in bacterial cells amorphous? A derivation from stress exposure experiments. Appl Microbiol Biot. 2019;103:1905–17.
CAS  Google Scholar 

48.
Steinbüchel A, Valentin HE. Diversity of bacterial polyhydroxyalkanoic acids. FEMS Microbiol Lett. 1995;128:219–28.
Google Scholar 

49.
Madison LL, Huisman GW. Metabolic engineering of poly(3-hydroxyalkanoates): From DNA to plastic. Microbiol Mol Biol Rev 1999;63:21–53.
CAS  PubMed  PubMed Central  Google Scholar 

50.
Kunioka M, Tamaki A, Doi Y. Crystalline and thermal properties of bacterial copolyesters: poly(3-hydroxybutyrate-co-3-hydroxyvalerate) and poly(3-hydroxybutyrate-co-4-hydroxybutyrate). Macromolecules. 1989;22:694–7.
CAS  Google Scholar 

51.
Doi Y, Kitamura S, Abe H. Microbial synthesis and characterization of poly(3-hydroxybutyrate-co-3-hydroxyhexanoate). Macromolecules. 1995;28:4822–8.
CAS  Google Scholar 

52.
Abe H, Doi Y. Side-chain effect of second monomer units on crystalline morphology, thermal properties, and enzymatic degradability for random copolyesters of (R)-3-hydroxybutyric acid with (R)-3-hydroxyalkanoic acids. Biomacromolecules. 2002;3:133–8.
CAS  PubMed  Google Scholar 

53.
Shimamura E, Kasuya K, Kobayashi G, Shiotani T, Shima Y, Doi Y. Physical-properties and biodegradability of microbial poly(3-hydroxybutyrate-co-3-hydroxyhexanoate). Macromolecules. 1994;27:878–80.
CAS  Google Scholar 

54.
Suriyamongkol P, Weselake R, Narine S, Moloney M, Shah S. Biotechnological approaches for the production of polyhydroxyalkanoates in microorganisms and plants—a review. Biotechnol Adv. 2007;25:148–75.
CAS  PubMed  Google Scholar 

55.
Rehm BHA, Kruger N, Steinbüchel A. A new metabolic link between fatty acid de novo synthesis and polyhydroxyalkanoic acid synthesis—the phaG gene from Pseudomonas putida KT2440 encodes a 3-hydroxyacyl-acyl carrier protein coenzyme A transferase. J Biol Chem. 1998;273:24044–51.
CAS  PubMed  Google Scholar 

56.
Rehm BHA, Mitsky TA, Steinbüchel A. Role of fatty acid de novo biosynthesis in polyhydroxyalkanoic acid (PHA) and rhamnolipid synthesis by pseudomonads: establishment of the transacylase (PhaG)-mediated pathway for PHA biosynthesis in Escherichia coli. Appl Environ Microbiol. 2001;67:3102–9.
CAS  PubMed  PubMed Central  Google Scholar 

57.
Fukui T, Shiomi N, Doi Y. Expression and characterization of (R)-specific enoyl coenzyme A hydratase involved in polyhydroxyalkanoate biosynthesis by Aeromonas caviae. J Bacteriol. 1998;180:667–73.
CAS  PubMed  PubMed Central  Google Scholar 

58.
Chek MF, Kim SY, Mori T, Arsad H, Samian MR, Sudesh K, et al. Structure of polyhydroxyalkanoate (PHA) synthase PhaC from Chromobacterium sp. USM2, producing biodegradable plastics. Sci Rep. 2017;7:5312. https://doi.org/10.1038/s41598-017-05509-4.
CAS  Article  PubMed  PubMed Central  Google Scholar 

59.
Mezzolla V, D’Urso OF, Poltronieri P. Role of PhaC type I and type II enzymes during PHA biosynthesis. Polymer. 2018;10:910. https://doi.org/10.3390/polym10080910.
CAS  Article  Google Scholar 

60.
Rehm BHA. Polyester synthases: natural catalysts for plastics. Biochem J. 2003;376:15–33.
CAS  PubMed  PubMed Central  Google Scholar 

61.
Tsuge T, Hyakutake M, Mizuno K. Class IV polyhydroxyalkanoate (PHA) synthases and PHA-producing Bacillus. Appl Microbiol Biotechnol. 2015;99:6231–40.
CAS  PubMed  Google Scholar 

62.
Lenz RW, Marchessault RH. Bacterial polyesters: biosynthesis, biodegradable plastics and biotechnology. Biomacromolecules. 2005;6:1–8.
CAS  PubMed  Google Scholar 

63.
Higuchi-Takeuchi M, Morisaki K, Numata K. A Screening method for the isolation of polyhydroxyalkanoate-producing purple non-sulfur photosynthetic bacteria from natural seawater. Front Microbiol. 2016;7:1509. https://doi.org/10.3389/fmicb.2016.01509.
Article  PubMed  PubMed Central  Google Scholar 

64.
Sabirova JS, Ferrer M, Regenhardt D, Timmis KN, Golyshin PN. Proteomic insights into metabolic adaptations in Alcanivorax borkumensis induced by alkane utilization. J Bacteriol. 2006;188:3763–73.
CAS  PubMed  PubMed Central  Google Scholar 

65.
Shi XL, Wu YH, Jin XB, Wang CS, Xu XW. Alteromonas lipolytica sp. nov., a poly-beta-hydroxybutyrateproducing bacterium isolated from surface seawater. Int J Syst Evol Microbiol. 2017;67:237–42.
CAS  PubMed  Google Scholar 

66.
Mohandas SP, Balan L, Jayanath G, Anoop BS, Philip R, Cubelio SS, et al. Biosynthesis and characterization of polyhydroxyalkanoate from marine Bacillus cereus MCCB 281 utilizing glycerol as carbon source. Int J Biol Macromol. 2018;119:380–92.
CAS  PubMed  Google Scholar 

67.
Sathiyanarayanan G, Saibaba G, Kiran GS, Selvin J. Process optimization and production of polyhydroxybutyrate using palm jaggery as economical carbon source by marine sponge-associated Bacillus licheniformis MSBN12. Bioproc. Biosyst Eng. 2013;36:1817–27.
CAS  Google Scholar 

68.
Lopez-Cortes A, Lanz-Landazuri A, Garcia-Maldonado JQ. Screening and isolation of PHB-producing bacteria in a polluted marine microbial mat. Microbiol Ecol. 2008;56:112–20.
CAS  Google Scholar 

69.
Prabhu NN, Santimano MC, Mavinkurve S, Bhosle SN, Garg S. Native granule associated short chain length polyhydroxyalkanoate synthase from a marine derived Bacillus sp. NQ-11/A2. Anton Leeuw. 2010;97:41–50.
CAS  Google Scholar 

70.
Hong JW, Song HS, Moon YM, Hong YG, Bhatia SK, Jung HR, et al. Polyhydroxybutyrate production in halophilic marine bacteria Vibrio proteolyticus isolated from the Korean peninsula. Bioprocess Biosyst Eng. 2019;42:603–10.
CAS  PubMed  Google Scholar 

71.
Kiran GS, Lipton AN, Priyadharshini S, Anitha K, Suarez LEC, Arasu MV, et al. Antiadhesive activity of poly-hydroxy butyrate biopolymer from a marine Brevibacterium casei MSI04 against shrimp pathogenic vibrios. Microb Cell Fact. 2014;13:114. https://doi.org/10.1186/s12934-014-0114-3.
CAS  Article  PubMed  PubMed Central  Google Scholar 

72.
Yamada M, Yukita A, Hanazumi Y, Yamahata Y, Moriya H, Miyazaki M, et al. Poly(3-hydroxybutyrate) production using mannitol as a sole carbon source by Burkholderia sp. AIU M5M02 isolated from a marine environment. Fish Sci. 2018;84:405–12.
CAS  Google Scholar 

73.
Methe BA, Nelson KE, Deming JW, Momen B, Melamud E, Zhang XJ, et al. The psychrophilic lifestyle as revealed by the genome sequence of Colwellia psychrerythraea 34H through genomic and proteomic analyses. Proc Natl Acid Sci USA. 2005;102:10913–8.
CAS  Google Scholar 

74.
Numata K, Morisaki K, Tomizawa S, Ohtani M, Demura T, Miyazaki M. Synthesis of poly- and oligo(hydroxyalkanoate)s by deep-sea bacteria, Colwellia spp., Moritella spp., and Shewanella spp. Polym J. 2013;45:1094–100.
CAS  Google Scholar 

75.
Hai T, Lange D, Rabus R, Steinbüchel A. Polyhydroxyalkanoate (PHA) accumulation in sulfate-reducing bacteria and identification of a class III PHA synthase (PhaEC) in Desulfococcus multivorans. Appl Environ Microbiol. 2004;70:4440–8.
CAS  PubMed  PubMed Central  Google Scholar 

76.
Wang H, Tomasch J, Jarek M, Wagner-Dobler I. A dual-species co-cultivation system to study the interactions between Roseobacters and dinoflagellates. Front Microbiol. 2014;5:311. https://doi.org/10.3389/fmicb.2014.00311.
Article  PubMed  PubMed Central  Google Scholar 

77.
Xiao N, Jiao NZ. Formation of polyhydroxyalkanoate in aerobic anoxygenic phototrophic bacteria and its relationship to carbon source and light availability. Appl Environ Microbiol. 2011;77:7445–50.
CAS  PubMed  PubMed Central  Google Scholar 

78.
Shrivastav A, Mishra SK, Shethia B, Pancha I, Jain D, Mishra S. Isolation of promising bacterial strains from soil and marine environment for polyhydroxyalkanoates (PHAs) production utilizing Jatropha biodiesel byproduct. Int J Biol Macromol. 2010;47:283–7.
CAS  PubMed  Google Scholar 

79.
Simon-Colin C, Raguénès G, Cozien J, Guezennec JG. Halomonas profundus sp. nov., a new PHA-producing bacterium isolated from a deep-sea hydrothermal vent shrimp. J Appl Microbiol. 2008;104:1425–32.
CAS  PubMed  Google Scholar 

80.
Lemechko P, Le Fellic M, Bruzaud S. Production of poly(3-hydroxybutyrate-co-3-hydroxyvalerate) using agro-industrial effluents with tunable proportion of 3-hydroxyvalerate monomer units. Int J Biol Macromol. 2019;128:429–34.
CAS  PubMed  Google Scholar 

81.
Han X, Satoh Y, Kuriki Y, Seino T, Fujita S, Suda T, et al. Polyhydroxyalkanoate production by a novel bacterium Massilia sp. UMI-21 isolated from seaweed, and molecular cloning of its polyhydroxyalkanoate synthase gene. J Biosci Bioeng. 2014;118:514–9.
CAS  PubMed  Google Scholar 

82.
Doronina NV, Trotsenko YA, Tourova TP. Methylarcula marina gen. nov., sp. nov. and Methylarcula terricola sp. nov.: novel aerobic, moderately halophilic, facultatively methylotrophic bacteria from coastal saline environments. Int J Syst Evol Microbiol. 2000;50:1849–59.
CAS  PubMed  Google Scholar 

83.
Liu XJ, Zhang J, Hong PH, Li ZJ. Microbial production and characterization of poly-3-hydroxybutyrate by Neptunomonas antarctica. Peerj. 2016;4:e2291. https://doi.org/10.7717/peerj.2291.
CAS  Article  PubMed  PubMed Central  Google Scholar 

84.
Cho JC, Giovannoni SJ. Oceanicola granulosus gen. nov., sp. nov. and Oceanicola batsensis sp. nov., poly-β-hydroxybutyrate-producing marine bacteria in the order ‘Rhodobacterales’. Int J Syst Evol Microbiol. 2004;54:1129–36.
CAS  PubMed  Google Scholar 

85.
Numata K, Morisaki K. Screening of marine bacteria to synthesize polyhydroxyalkanoate from lignin: contribution of lignin derivatives to biosynthesis by Oceanimonas doudoroffii. ACS Sustain Chem Eng. 2015;3:569–73.
CAS  Google Scholar 

86.
Boyandin AN, Kalacheva GS, Rodicheva EK, Volova TG. Synthesis of reserve polyhydroxyalkanoates by luminescent bacteria. Microbiology. 2008;77:318–23.
CAS  Google Scholar 

87.
Wang Q, Zhang HX, Chen Q, Chen XL, Zhang YZ, Qi QS. A marine bacterium accumulates polyhydroxyalkanoate consisting of mainly 3-hydroxydodecanoate and 3-hydroxydecanoate. World J Microbiol Biotechnol. 2010;26:1149–53.
Google Scholar 

88.
Simon-Colin C, Alain K, Colin S, Cozien J, Costa B, Guezennec JG, et al. A novel mcl PHA-producing bacterium, Pseudomonas guezennei sp. nov., isolated from a ‘kopara’ mat located in Rangiroa, an atoll of French Polynesia. J Appl Microbiol. 2008;104:581–6.
CAS  PubMed  Google Scholar 

89.
Jamil N, Ahmed N, Edwards DH. Characterization of biopolymer produced by Pseudomonas sp. CMG607w of marine origin. J Gen Appl Microbiol. 2007;53:105–9.
CAS  PubMed  Google Scholar 

90.
Higuchi-Takeuchi M, Numata K. Acetate-inducing metabolic states enhance polyhydroxyalkanoate production in marine purple non-sulfur bacteria under aerobic conditions. Front Bioeng Biotechnol. 2019;7:118. https://doi.org/10.3389/fbioe.2019.00118.
Article  PubMed  PubMed Central  Google Scholar 

91.
Higuchi-Takeuchi M, Morisaki K, Toyooka K, Numata K. Synthesis of high-molecular-weight polyhydroxyalkanoates by marine photosynthetic purple bacteria. PLoS ONE. 2016;11. https://doi.org/10.1371/journal.pone.0160981.

92.
Gonzalez-Garcia Y, Nungaray J, Cordova J, Gonzalez-Reynoso O, Koller M, Atlic A, et al. Biosynthesis and characterization of polyhydroxyalkanoates in the polysaccharide-degrading marine bacterium Saccharophagus degradans ATCC 43961. J Ind Microbiol Biotechnol. 2008;35:629–33.
CAS  PubMed  Google Scholar 

93.
Ting L, Williams TJ, Cowley MJ, Lauro FM, Guilhaus M, Raftery MJ, et al. Cold adaptation in the marine bacterium, Sphingopyxis alaskensis, assessed using quantitative proteomics. Environ Microbiol. 2010;12:2658–76.
CAS  PubMed  Google Scholar 

94.
Shrivastav A, Mishra SK, Mishra S. Polyhydroxyalkanoate (PHA) synthesis by Spirulina subsalsa from Gujarat coast of India. Int J Biol Macromol. 2010;46:255–60.
CAS  PubMed  Google Scholar 

95.
Sasidharan RS, Bhat SG, Chandrasekaran M. Biocompatible polyhydroxybutyrate (PHB) production by marine Vibrio azureus BTKB33 under submerged fermentation. Ann Microbiol. 2015;65:455–65.
CAS  Google Scholar 

96.
Mohandas SP, Balan L, Lekshmi N, Cubelio SS, Philip R, Bright SIS. Production and characterization of polyhydroxybutyrate from Vibrio harveyi MCCB 284 utilizing glycerol as carbon source. J Appl Microbiol. 2017;122:698–707.
CAS  PubMed  Google Scholar 

97.
Wei YH, Chen WC, Wu HS, Janarthanan OM. Biodegradable and biocompatible biomaterial, polyhydroxybutyrate, produced by an indigenous Vibrio sp. BM-1 isolated from marine environment. Mar Drugs. 2011;9:615–24.
CAS  PubMed  PubMed Central  Google Scholar 

98.
Numata K, Doi Y. Biosynthesis of Polyhydroxyalkanaotes by a novel facultatively anaerobic Vibrio sp. under marine conditions. Mar Biotechnol. 2012;14:323–31.
CAS  PubMed  Google Scholar 

99.
Stabili L, Cardone F, Alifano P, Tredici SM, Piraino S, Corriero G, et al. Epidemic Mortality of the sponge Ircinia variabilis (Schmidt, 1862) associated to proliferation of a Vibrio Bacterium. Microb Ecol. 2012;64:802–13.
PubMed  Google Scholar 

100.
Foong CP, Lau NS, Deguchi S, Toyofuku T, Taylor TD, Sudesh K, et al. Whole genome amplification approach reveals novel polyhydroxyalkanoate synthases (PhaCs) from Japan Trench and Nankai Trough seawater. BMC Microbiol. 2014;14:318. https://doi.org/10.1186/s12866-014-0318-z.
CAS  Article  PubMed  PubMed Central  Google Scholar 

101.
Doi Y, Kanesawa Y, Tanahashi N, Kumagai Y. Biodegradation of microbial polyesters in the marine-environment. Polym Degrad Stabil. 1992;36:173–7.
CAS  Google Scholar 

102.
Tsuji H, Suzuyoshi K. Environmental degradation of biodegradable polyesters 2. Poly(ε-caprolactone), poly[(R)-3-hydroxybutyrate], and poly(L-lactide) films in natural dynamic seawater. Polym Degrad Stabil. 2002;75:357–65.
CAS  Google Scholar 

103.
Tsuji H, Suzuyoshi K. Environmental degradation of biodegradable polyesters 1. Poly(ε-caprolactone), poly[(R)-3-hydroxybutyrate], and poly(L-lactide) films in controlled static seawater. Polym Degrad Stabil. 2002;75:347–55.
CAS  Google Scholar 

104.
Rutkowska M, Krasowska K, Heimowska A, Adamus G, Sobota M, Musiol M, et al. Environmental degradation of blends of atactic poly (R,S)-3-hydroxybutyrate with natural PHBV in Baltic sea water and compost with activated sludge. J Polym Environ. 2008;16:183–91.
CAS  Google Scholar 

105.
Volova TG, Boyandin AN, Vasil’ev AD, Karpov VA, Kozhevnikov IV, Prudnikova SV, et al. Biodegradation of polyhydroxyalkanoates (PHAs) in the South China Sea and identification of PHA-degrading bacteria. Microbiology. 2011;80:252. https://doi.org/10.1134/S0026261711020184.
CAS  Article  Google Scholar 

106.
Thellen C, Coyne M, Froio D, Auerbach M, Wirsen C, Ratto JA. A processing, characterization and marine biodegradation study of melt-extruded polyhydroxyalkanoate (PHA) films. J Polym Environ. 2008;16:1–11.
CAS  Google Scholar 

107.
Deroine M, Cesar G, Le Duigou A, Davies P, Bruzaud S. Natural degradation and biodegradation of poly(3-hydroxybutyrate-co-3-hydroxyvalerate) in liquid and solid marine environments. J Polym Environ. 2015;23:493–505.
CAS  Google Scholar 

108.
Imam SH, Gordon SH, Shogren RL, Tosteson TR, Govind NS, Greene RV. Degradation of starch-poly(beta-hydroxybutyrate-co-beta-hydroxyvalerate) bioplastic in tropical coastal waters. Appl Environ Microbiol. 1999;65:431–7.
CAS  PubMed  PubMed Central  Google Scholar 

109.
Suzuki M, Tachibana Y, Kazahaya J, Takizawa R, Muroi F, Kasuya K. Difference in environmental degradability between poly(ethylene succinate) and poly(3-hydroxybutyrate). J Polym Res. 2017;24:217. https://doi.org/10.1007/s10965-017-1383-4.
CAS  Article  Google Scholar 

110.
Dilkes-Hoffman LS, Lant PA, Laycock B, Pratt S. The rate of biodegradation of PHA bioplastics in the marine environment: a meta-study. Mar Pollut Bull. 2019;142:15–24.
CAS  PubMed  Google Scholar 

111.
Zettler ER, Mincer TJ, Amaral-Zettler LA. Life in the “Plastisphere”: microbial communities on plastic marine debris. Environ Sci Technol. 2013;47:7137–46.
CAS  PubMed  Google Scholar 

112.
Morohoshi T, Ogata K, Okura T, Sato S. Molecular characterization of the bacterial community in biofilms for degradation of poly(3-hydroxybutyrate-co-3-hydroxyhexanoate) films in seawater. Microbes Environ. 2018;33:19–25.
PubMed  PubMed Central  Google Scholar 

113.
Pinnell LJ, Turner JW. Shotgun metagenomics reveals the benthic microbial community response to plastic and bioplastic in a coastal marine environment. Front Microbiol. 2019;10:1252. https://doi.org/10.3389/fmicb.2019.01252.
Article  PubMed  PubMed Central  Google Scholar 

114.
Zadjelovic V, Chhun A, Quareshy M, Silvano E, Hernandez-Fernaud JR, Aguilo-Ferretjans MM, et al. Beyond oil degradation: enzymatic potential of Alcanivorax to degrade natural and synthetic polyesters. Environ Microbiol. 2020;22:1356–69.
CAS  PubMed  PubMed Central  Google Scholar 

115.
Kita K, Ishimaru K, Teraoka M, Yanase H, Kato N. Properties of poly(3-hydroxybutyrate) depolymerase from a marine bacterium, Alcaligenes faecalis AE122. Appl Environ Microbiol. 1995;61:1727–30.
CAS  PubMed  PubMed Central  Google Scholar 

116.
Mergaert J, Wouters A, Anderson C, Swings J. In-situ biodegradation of poly(3-hydroxybutyrate) and poly(3-hydroxybutyrate-co-3-hydroxyvalerate) in natural-waters. Can J Microbiol. 1995;41:154–9.
CAS  PubMed  Google Scholar 

117.
Kato C, Honma A, Sato S, Okura T, Fukuda R, Nogi Y. Poly 3-hydroxybutyrate-co-3-hydroxyhexanoate films can be degraded by the deep-sea microbes at high pressure and low temperature conditions. High Press Res. 2019;39:248–57.
CAS  Google Scholar 

118.
Volova TG, Boyandin AN, Vasiliev AD, Karpov VA, Prudnikova SV, Mishukova OV, et al. Biodegradation of polyhydroxyalkanoates (PHAs) in tropical coastal waters and identification of PHA-degrading bacteria. Polym Degrad Stabil. 2010;95:2350–9.
CAS  Google Scholar 

119.
Ma WT, Lin JH, Chen HJ, Chen SY, Shaw GC. Identification and characterization of a novel class of extracellular poly(3-hydroxybutyrate) depolymerase from Bacillus sp Strain NRRL B-14911. Appl Environ Microbiol. 2011;77:7924–32.
CAS  PubMed  PubMed Central  Google Scholar 

120.
Kasuya K, Mitomo H, Nakahara M, Akiba A, Kudo T, Doi Y. Identification of a marine benthic P(3HB)-degrading bacterium isolate and characterization of its P(3HB) depolymerase. Biomacromolecules. 2000;1:194–201.
CAS  PubMed  Google Scholar 

121.
Ghanem NB, Mabrouk MES, Sabry SA, El-Badan DES. Degradation of polyesters by a novel marine Nocardiopsis aegyptia sp. nov.: Application of Plackett-Burman experimental design for the improvement of PHB depolymerase activity. J Gen Appl Microbiol. 2005;51:151–8.
CAS  PubMed  Google Scholar 

122.
Leathers TD, Govind NS, Greene RV. Biodegradation of poly(3-hydroxybutyrate-co-3-hydroxyvalerate) by a tropical marine bacterium, Pseudoalteromonas sp. NRRL b-30083. J Polym Environ. 2000;8:119–24.
CAS  Google Scholar 

123.
Uefuji M, Kasuya K, Doi Y. Enzymatic degradation of poly (R)-3-hydroxybutyrate: secretion and properties of PHB depolymerase from Pseudomonas stutzeri. Polym Degrad Stabil. 1997;58:275–81.
CAS  Google Scholar 

124.
Sung CC, Tachibana Y, Suzuki M, Hsieh WC, Kasuya K. Identification of a poly(3-hydroxybutyrate)-degrading bacterium isolated from coastal seawater in Japan as Shewanella sp. Polym Degrad Stabil. 2016;129:268–74.
CAS  Google Scholar 

125.
Mabrouk MM, Sabry SA. Degradation of poly (3-hydroxybutyrate) and its copolymer poly (3-hydroxybutyrate-co-3-hydroxyvalerate) by a marine Streptomyces sp. SNG9. Microbiol Res. 2001;156:323–35.
CAS  PubMed  Google Scholar 

126.
Kita K, Mashiba S, Nagita M, Ishimaru K, Okamoto K, Yanase H, et al. Cloning of poly(3-hydroxybutyrate) depolymerase from a marine bacterium, Alcaligenes faecalis AE122, and characterization of its gene product. BBA-Gene Struct Expr. 1997;1352:113–22.
CAS  Google Scholar 

127.
Kasuya K, Takano T, Tezuka Y, Hsieh WC, Mitomo H, Doi Y. Cloning, expression and characterization of a poly(3-hydroxybutyrate) depolymerase from Marinobacter sp. NK-1. Int J Biol Macromol. 2003;33:221–6.
CAS  PubMed  Google Scholar 

128.
Ohura T, Kasuya K, Doi Y. Cloning and characterization of the polyhydroxybutyrate depolymerase gene of Pseudomonas stutzeri and analysis of the function of substrate-binding domains. Appl Environ Microbiol. 1999;65:189–97.
CAS  PubMed  PubMed Central  Google Scholar 

129.
Hisano T, Kasuya K, Tezuka Y, Ishii N, Kobayashi T, Shiraki M, et al. The crystal structure of polyhydroxybutyrate depolymerase from Penicillium funiculosum provides insights into the recognition and degradation of biopolyesters. J Mol Biol. 2006;356:993–1004.
CAS  PubMed  Google Scholar 

130.
Kasuya K, Inoue Y, Tanaka T, Akehata T, Iwata T, Fukui T, et al. Biochemical and molecular characterization of the polyhydroxybutyrate depolymerase of Comamonas acidovorans YM1609, isolated from freshwater. Appl Environ Microbiol. 1997;63:4844–52.
CAS  PubMed  PubMed Central  Google Scholar 

131.
Sung CC, Tachibana Y, Kasuya K. Characterization of a thermolabile poly(3-hydroxybutyrate) depolymerase from the marine bacterium Shewanella sp. JKCM-AJ-6,1α. Polym Degrad Stabil. 2016;129:212–21.
CAS  Google Scholar 

132.
NOAA. Sea surface temperature (sst) contour charts. 2020. https://www.ospo.noaa.gov/Products/ocean/sst/contour/.

133.
Kroeker KJ, Kordas RL, Crim R, Hendriks IE, Ramajo L, Singh GS, et al. Impacts of ocean acidification on marine organisms: quantifying sensitivities and interaction with warming. Glob Change Biol. 2013;19:1884–96.
Google Scholar 

134.
Xue DW, Zhang XQ, Lu XL, Chen G, Chen ZH. Molecular and evolutionary mechanisms of cuticular wax for plant drought tolerance. Front Plant Sci. 2017;8:621. https://doi.org/10.3389/fpls.2017.00621.
Article  PubMed  PubMed Central  Google Scholar 

135.
Yeats TH, Rose JKC. The formation and function of plant cuticles. Plant Physiol. 2013;163:5–20.
CAS  PubMed  PubMed Central  Google Scholar 

136.
Philippe G, Sørensen I, Jiao C, Sun X, Fei Z, Domozych DS, et al. Cutin and suberin: assembly and origins of specialized lipidic cell wall scaffolds. Curr Opin Plant Biol. 2020;55:11–20.
CAS  PubMed  Google Scholar 

137.
Broder L, Tesi T, Salvado JA, Semiletov IP, Dudarev OV, Gustafsson O. Fate of terrigenous organic matter across the Laptev Sea from the mouth of the Lena River to the deep sea of the Arctic interior. Biogeosciences. 2016;13:5003–19.
Google Scholar 

138.
Broder L, Tesi T, Andersson A, Eglinton TI, Semiletov IP, Dudarev OV, et al. Historical records of organic matter supply and degradation status in the East Siberian Sea. Org Geochem. 2016;91:16–30.
Google Scholar 

139.
Goni MA, Yunker MB, Macdonald RW, Eglinton TI. Distribution and sources of organic biomarkers in arctic sediments from the Mackenzie River and Beaufort Shelf. Mar Chem. 2000;71:23–51.
CAS  Google Scholar 

140.
Prahl FG, Ertel JR, Goni MA, Sparrow MA, Eversmeyer B. Terrestrial organic-carbon contributions to sediments on the Washington margin. Geochim Cosmochim Acta. 1994;58:3035–48.
CAS  Google Scholar 

141.
Kaal J, Serrano O, Cortizas AM, Baldock JA, Lavery PS. Millennial-scale changes in the molecular composition of Posidonia australis seagrass deposits: Implications for Blue Carbon sequestration. Org Geochem. 2019;137:103898. https://doi.org/10.1016/j.orggeochem.2019.07.007.
CAS  Article  Google Scholar 

142.
Moran KL, Bjorndal KA. Simulated green turtle grazing affects nutrient composition of the seagrass Thalassia testudinum. Mar Biol. 2007;150:1083–92.
CAS  Google Scholar 

143.
Olsen JL, Rouze P, Verhelst B, Lin YC, Bayer T, Collen J, et al. The genome of the seagrass Zostera marina reveals angiosperm adaptation to the sea. Nature. 2016;530:331–5.
CAS  PubMed  Google Scholar 

144.
Sugiura H, Kawasaki Y, Suzuki T, Maegawa M. The structural and histochemical analyses and chemical characters of the cuticle and epidermal walls of cotyledon in ungerminated seeds of Zostera marina L. Fish Sci. 2009;75:369–77.
CAS  Google Scholar 

145.
Lu B, Wang GX, Huang D, Ren ZL, Wang XW, Zhen ZC, et al. Comparison of PCL degradation in different aquatic environments: effects of bacteria and inorganic salts. Polym Degrad Stabil. 2018;150:133–9.
CAS  Google Scholar 

146.
Molitor R, Bollinger A, Kubicki S, Loeschcke A, Jaeger KE, Thies S. Agar plate-based screening methods for the identification of polyester hydrolysis by Pseudomonas species. Microb Biotechnol. 2020;13:274–84.
CAS  PubMed  Google Scholar 

147.
Suzuki M, Tachibana Y, Oba K, Takizawa R, Kasuya K. Microbial degradation of poly(ε-caprolactone) in a coastal environment. Polym Degrad Stabil. 2018;149:1–8.
CAS  Google Scholar 

148.
Almeida EL, AFC Rincon, Jackson SA, ADW Dobson. In silico screening and heterologous expression of a polyethylene terephthalate hydrolase (PETase)-like enzyme (SM14est) with polycaprolactone (PCL)-degrading activity, from the marine sponge-derived strain Streptomyces sp. SM14. Front Microbiol. 2019;10:2187. https://doi.org/10.3389/fmicb.2019.02187.
Article  PubMed  Google Scholar 

149.
Sekiguchi T, Sato T, Enoki M, Kanehiro H, Uematsu K, Kato C. Isolation and characterization of biodegradable plastic degrading bacteria from deep-sea environments. JAMSTEC Rep. Res Dev. 2011;11:33–41.
Google Scholar 

150.
Kovacic F, Babic N, Krauss U, Jaeger K. Classification of lipolytic enzymes from bacteria. In: Handbook of hydrocarbon and lipid microbiology. Cham: Springer; 2019;24. https://doi.org/10.1007/978-3-319-39782-5_39-1.

151.
Arpigny JL, Jaeger KE. Bacterial lipolytic enzymes: classification and properties. Biochem J. 1999;343:177–83.
CAS  PubMed  PubMed Central  Google Scholar 

152.
Nikolaivits E, Kanelli M, Dimarogona M, Topakas E. A middle-aged enzyme still in its prime: recent advances in the field of cutinases. Catalysts. 2018;8:612. https://doi.org/10.3390/catal8120612.
CAS  Article  Google Scholar 

153.
Chen S, Su LQ, Chen J, Wu J. Cutinase: characteristics, preparation, and application. Biotechnol Adv. 2013;31:1754–67.
CAS  PubMed  Google Scholar 

154.
Chen S, Tong X, Woodard RW, Du GC, Wu J, Chen J. Identification and characterization of bacterial cutinase. J Biol Chem. 2008;283:25854–62.
CAS  PubMed  PubMed Central  Google Scholar 

155.
Reis P, Holmberg K, Watzke H, Leser ME, Miller R. Lipases at interfaces: a review. Adv Colloid Interfac. 2009;147:237–50.
Google Scholar 

156.
Zumstein MT, Rechsteiner D, Roduner N, Perz V, Ribitsch D, Guebitz GM, et al. Enzymatic hydrolysis of polyester thin films at the nanoscale: effects of polyester structure and enzyme active-site accessibility. Environ Sci Technol. 2017;51:7476–85.
CAS  PubMed  Google Scholar 

157.
Shi K, Jing J, Song L, Su TT, Wang ZY. Enzymatic hydrolysis of polyester: Degradation of poly(ε-caprolactone) by Candida antarctica lipase and Fusarium solani cutinase. Int J Biol Macromol. 2020;144:183–9.
CAS  PubMed  Google Scholar 

158.
Trodler P, Pleiss J. Modeling structure and flexibility of Candida antarctica lipase B in organic solvents. BMC Struct Biol. 2008;8:9. https://doi.org/10.1186/1472-6807-8-9.
CAS  Article  PubMed  PubMed Central  Google Scholar 

159.
Hajighasemi M, Nocek BP, Tchigvintsev A, Brown G, Flick R, Xu XH, et al. Biochemical and structural insights into enzymatic depolymerization of polylactic acid and other polyesters by microbial carboxylesterases. Biomacromolecules. 2016;17:2027–39.
CAS  PubMed  PubMed Central  Google Scholar 

160.
Danso D, Schmeisser C, Chow J, Zimmermann W, Wei R, Leggewie C, et al. New insights into the function and global distribution of polyethylene terephthalate (PET)-degrading bacteria and enzymes in marine and terrestrial metagenomes. Appl Environ Microbiol. 2018;84:e02773–17. https://doi.org/10.1128/AEM.02773-17.
CAS  PubMed  PubMed Central  Google Scholar 

161.
Bollinger A, Thies S, Knieps-Grunhagen E, Gertzen C, Kobus S, Hoppner A, et al. A novel polyester hydrolase from the marine bacterium Pseudomonas aestusnigri – Structural and functional insights. Front Microbiol. 2020;11:114. https://doi.org/110.3389/fmicb.2020.00114.
PubMed  PubMed Central  Google Scholar 

162.
Shinomiya M, Iwata T, Kasuya K, Doi Y. Cloning of the gene for poly(3-hydroxybutyric acid) depolymerase of Comamonas testosteroni and functional analysis of its substrate-binding domain. FEMS Microbiol Lett. 1997;154:89–94.
CAS  PubMed  Google Scholar  More