Introduction

The Caryophyllales Network — This specialists' network was initiated in 2011 (Arias & al. 2016) with the aim to mobilize and coordinate the international endeavour to assemble a “Global synthesis of species diversity in the angiosperm order Caryophyllales” (Borsch & al. 2015). A networked and informatics-supported approach is taken to build up baseline information on species in a single open online resource. Providing open access to that information will facilitate and inspire research, avoid duplication of effort, constantly improve quality and consistency of the underlying data, stimulate contributions and directly contribute to other global initiatives. Concurrently, open research questions in biodiversity informatics but also evolution and taxonomy become obvious and can be tackled. This already goes beyond the call for online floras issued by Brach & Boufford (2011). The first result of this collaborative effort was the publication of a family and genus-level checklist based on the latest (published) phylogenetic knowledge (Hernández-Ledesma & al. 2015), which has since been updated frequently on the Caryophyllales website (Caryophyllales Network 2015+).

The database, online portal and publication pipelines of the Caryophyllales Network are using the EDIT Platform for Cybertaxonomy (Ciardelli & al. 2009; Berendsohn 2010) to capture, process, maintain and publish the data. Short-term aims in the assembly of information are to document the state of the art with respect to:

Medium and long term aims:

The Caryophyllales initiative explicitly aims at integrating published results and data, ongoing research, and usage of the information, not merely at aggregating data. Its main target audience is researchers, although it will also serve as a quality source for a broad range of user communities, either directly or by contributing to aggregators such as the Global Biodiversity Information Facility (GBIF 2004+), Catalogue of Life (CoL; Bisby 1997; Roskov & al. 2018), Encyclopedia of Life (EoL; Parr & al. 2014) and the World Flora Online (WFO; Wyse Jackson & Miller 2015) initiative.

The World Flora Online initiative — (Wyse Jackson & Miller 2015; Loizeau & Wyse Jackson 2017). The World Flora Online will be an open-access, web-based compendium of the world's plant species. This is answering Target 1 of the updated Global Strategy for Plant Conservation (GSPC) 2011–2020 of the U.N. Convention on Biological Diversity (CBD; COP 2010; Sharrock 2012). In response to this target, an international consortium of institutions and organizations was formed to collaborate on providing the content. The 11th Conference of the Parties to the Convention on Biological Diversity adopted a decision welcoming the World Flora Online initiative (CBD-SBSTTA 2012). Since January 2013, up to June 2018, 41 institutions have signed a Memorandum of Understanding, agreeing to their participation in the World Flora Online initiative.

WFO will assemble this compendium by aggregating information from monographs, taxonomic networks and floras based on a consensus “taxonomic backbone”, i.e. a list of taxa with their synonyms. The initial backbone is formed by the checklist assembled by a consortium of institutions under the lead of Missouri Botanical Garden and Royal Botanic Gardens, Kew as the outcome of Target 1 of the GSPC 2001–2010 (COP 2002), namely “The Plant List” (TPL 2013). A unique WFO Identifier has been assigned to each of the names in the backbone and will be maintained permanently. Presently, WFO's pragmatic approach to aggregation will be name-based, i.e. the incoming information will be linked to the backbone taxon to which the incoming name belongs, thus for the time being ignoring possible concept conflicts (see below). The Caryophyllales Network was recognized by the WFO Council as one of WFO's TENs (Taxonomic Expert Networks), whose responsibility it is to take on a specific “taxonomic slice” of the WFO and revise the taxonomy. In this context, one of the first aims for the Network is to provide an updated WFO backbone for Caryophyllales, which implies that all names in the WFO backbone derived from The Plant List have to be resolved in the Caryophyllales database, and are therefore to be placed within the best currently available taxonomic treatment for the respective species, genus or family.

Why Nepenthes? — The assembly of the taxonomic backbone is a tedious task, given the number of unresolved names that have been introduced into The Plant List and therefore into the original WFO backbone. In 2016, we were looking for a taxonomic group to try out the necessary procedures and, if possible, go beyond mere backbone data. We were therefore in need of a relatively small family with clear monophyly also at the generic level, and with existing, high-quality, electronic, open, taxonomic data that we could use as a starting point. The genus Nepenthes is clearly monophyletic (Meimberg & Heubl 2006; Meimberg & al. 2006). With most of the taxa in the monogeneric Nepenthaceae occurring in the Flora Malesiana area and therefore already present in the EDIT Platform, this family appeared to be an ideal candidate. Thus commenced the first author's dive into the exciting world of pitcher plants and their protagonists around the world.

Material and methods

Sources — The most important single source of data for the treatment presented here was the Nepenthaceae treatment in Flora Malesiana (Cheek & Jebb 2001), which was parsed and imported into an EDIT Platform database between 2010 and 2015 by Thomas Hamann and Andreas Müller (Roos & al. 2012; Hamann & al. 2014). Unfortunately, the project supporting this process ended prematurely and no further resources could be committed up to now to iron out the remaining problems in the online version of Flora Malesiana ( http://portal.cybertaxonomy.org/flora-malesiana/). The Nepenthes treatment was extracted from the Flora Malesiana database and imported into the Caryophyllales database. Further data-cleaning was effected to correct problems remaining from the import, and the extra-Malesian taxa were added, initially using the “skeletal revision” of the genus by Jebb & Cheek (1997). New taxa were added after reviewing many taxonomic treatments, which cannot be listed in full here (they appear as sources in the online version); the online Wikipedia pages for Nepenthes proved to be important pointers to new publications. Further sources that should be mentioned here include the entertaining books by McPherson & al. (McPherson 2009, 2011) and the treatments by Clarke (1997, 2001), unfortunately with a number of newly described species behind a print-only paywall and thus raising an unnecessary obstacle to reusing the data. The travel diary of Catalano (2010) from Thailand should also be cited; he gracefully published the protologues and articles with further species descriptions from Indo-China online on his website ( http://www.marcellocatalano.com). The ongoing publication of contributions to a full revision of the family by Cheek & Jebb (see Cheek & al. 2018 for a list of references) are an important source, also providing some data on the subgeneric classification for the species, mostly formulated as informal species groups. With regard to species-level phylogenetic analyses, Brittnacher (2011) gave an overview of developments up to that year, citing 78 species sampled, and Schwallier & al. 2016 (five species) and Bunawan & al. 2017 (11 species) are apparently the most recently published analyses. It is noteworthy that the available molecular phylogenetic investigations on Nepenthes (e.g. Meimberg & Heubl 2006; Meimberg & al. 2006; Bunawal & al. 2017) employ few genomic regions and sample mostly only a single individual per species. The evolutionary testing of species limits and species relationships is therefore just at the beginning. Clarke & al. (2018) summarized the results and also provided a checklist of species (stating the caveat that “even the four authors of this article cannot agree upon the status of approximately 10% of Nepenthes species”), reviewed the taxonomic groups established by, and since the revision of, Danser (1928), and critically reviewed the “taxonomic inflation” caused by a recent flurry of new species publications.

The Name Matching Tool of the Tropicos website (Tropicos 1991+) was valuable for detecting discrepancies in nomenclatural reference citations. The slight taxonomic complication of having so many present-day carnivorous plant enthusiasts discussing species limits etc. was insignificant compared to the problem of sorting out names coined by horticulturists in the second half of the 19th century. Periodicals like the Gardeners' Chronicle contain numerous names, often of artificial hybrids, which have to be considered in a nomenclatural context. Many of these are contained in the International Plant Names Index (IPNI 2004+), sometimes including names attributed to later (nomenclaturally superfluous) publications. From there, they also found their way into The Plant List (TPL 2013) and consequently into the taxonomic backbone of the World Flora Online database. This meant that they had to be treated without exception, and in processing those names additional ones emerged from that type of horticultural literature. That these periodicals are almost completely available online, mostly in the Biodiversity Heritage Library (BHL 2005+), greatly facilitated the investigation. Protologues were also located using JSTOR (2000+), Google Scholar and results obtained using search engines.

Tools: the EDIT Platform — The EDIT Platform for Cybertaxonomy (Ciardelli & al. 2009; Berendsohn 2010; Kohlbecker 2017) provides a comprehensive set of tools for editing and management of taxonomic data (individually or collaboratively), fully customizable online access to that data, and other means of data publication and data exchange (BGBM 2011+). The development of the Platform predates the EU-funded EDIT (European Distributed Institute of Taxonomy) project, which provided major funding; the Platform is based on information models and software developed starting in the early 1990s (Berendsohn & al. 2003). At the core of the platform is the EDIT Common Data Model (CDM), a comprehensive information model covering the entire information domain: names and classifications, descriptive data (morphological, molecular), multimedia, geographic information, literature, specimens, people and user-defined content items (Anonymous 2008+; Müller & al. 2017). The open-source software provides solid and open interfaces including a Java programmers' library and a Web Service Layer. The most important applications are the highly configurable Taxonomic Editor (TaxEditor) for data entry and manipulation, the CDM Data Portal (Kohlbecker & al. 2008) including geo-services (Roca & al. 2009) for online display, and the Platform Web Services (Güntsch & al. 2017a). The TaxEditor (available for Windows, Linux and Macintosh operating systems) is the local client that allows users to enter, edit, export and otherwise manipulate the data in the CDM database, which resides on a CDM Server connected to the Internet. Multiple authors can work simultaneously on the same database. The Caryophyllales database is hosted by the BGBM. Data in Platform databases can be exported into various formats. In the context of the work presented here, we used the CDM-light export to generate tables for further processing. The EDIT Platform is used by a number of initiatives; see BGBM (2016+) for a list of publicly accessible Data Portals. The role of the EDIT Platform in the context of the Caryophyllales synthesis is more fully detailed in Borsch & al. (2015).

Taxonomic concepts and source citations in the EDIT Platform — To understand the Platform's handling of taxonomic concepts, a brief clarification of terminology is required.

The Classification is the uppermost hierarchical element in the Platform's handling of taxa. Several classifications can reside in a single database. This is useful, for example, when there are alternative views on taxon circumscriptions (e.g. in the treatment of the genera Hieracium and Pilosella in the Cichorieae, see Kilian & al. 2009+). An accepted name (“correct name” according to the rules of nomenclature) nested within a classification designates a Taxon Node, representing a taxon in a given classification. If a taxon node is assigned to a taxon of a higher rank, the latter is referred to as the parent taxon, the former as the latter's child taxon.

A Taxon is a taxonomic group with the data that define its circumscription and describe its properties. The circumscription of the taxon is indicated by means of a circumscription or concept reference (“sec.-” or “secundum-” reference, Berendsohn 1995, 1997; Berendsohn & Geoffroy 2007), normally a bibliographic reference clarifying the distinction of this taxon from other taxa. One and the same taxon may occur in several classifications, but it is also possible that two different taxa (taxon concepts) carry the same name in separate classifications. In themselves, classifications should be taxonomically consistent, i.e. every name should only occur once (as a taxon name or a synonym) in a given classification (except when cited as a misapplied name or pro-parte synonym).

Scientific names (as well as the names given to pseudotaxa such as unnamed clades) are assigned to taxa, synonyms or misapplied names. Ranks of names follow the hierarchy defined in the nomenclatural codes. However, the user is free to insert further ranks at any place in the tree thus forming new or mixed hierarchical levels.

The EDIT Platform offers various ways to indicate the source of a taxonomic concept. In addition to the “sec. reference” that indicates the circumscription of the taxon, the assignment of synonyms or of misapplied names may carry an individual “sec. reference”, too, here indicating who has linked this name to the concept of the taxon. This allows addition of such information without having to create a new taxon concept. In case of misapplied names, this attribution is distinguished from the “sensu” reference that indicates the actual misapplication. Moreover, factual data (as virtually all data items in the Platform) may be individually referenced (including the name used in the source), e.g. geographic distribution, ecological data, or uses of the taxon. It is also possible to assign a block of sourced factual data (e.g. the treatment in a Flora) to the taxon. Notes (or otherwise designated free-text fields) can be used to indicate the state of revision of the concept. These features were all used in the Nepenthaceae database presented here.

The Platform also allows linkage of different taxonomic concepts. An example: Nepenthes dubia Danser was treated by Jebb & Cheek (1997) and Cheek & Jebb (2001) as a taxon including N. tenuis Nerz & Wistuba; a hypothesis that was rejected by Clarke (2001), McPherson (2009) and Clarke & al. 2018, who treat N. tenuis as a separate taxon. In the primary classification in the database (visible in the portal), we presently follow Clarke's opinion, so we do have a taxon N. dubia sec. Clarke (2001), but we maintain the Flora Malesiana treatment as a taxon N. dubia sec. Cheek & Jebb (2001) in a parallel classification (for technical reasons not yet shown in the portal). The EDIT Platform allows us to establish a concept relationship (Gradstein & al. 2001) between the two Taxon Nodes, in this case “N. dubia sec. Cheek & Jebb (2001) includes, and is not congruent with, N. dubia sec. Clarke (2001)”.

Specimens in the EDIT Platform — The Platform covers specimens in a comprehensive way. The data items available in the ABCD (Access to Biological Collection Data; Berendsohn 2007–2010; Holetschek & al. 2012) and DwC (Darwin Core; Wieczorek & al. 2009) standards are included in the Platform, such as those relating to the collection site, collection event, processing, herbarium location, and identification history of the specimen. Moreover, the hierarchical modelling of such data in the CDM allows linkage of all derivatives of materials collected, i.e. specimen duplicates, images taken at the collection site, images of specimens, tissue samples, DNA samples, etc. (Kilian & al. 2015; Plitzner & al. 2017, 2017a).

Descriptive information in the EDIT Platform — The hierarchical structure for specimens is also important in the context of descriptive information. Aggregation of atomized character data from specimens is a current research topic in biodiversity informatics (Plitzner & al. 2017, 2017a; Henning & al. 2018). The CDM already fully supports the comprehensive TDWG Standard SDD (Structured Descriptive Data; Hagedorn & al. 2005), and linkage to ontologies or terminology services (Karam & al. 2016) are features under development, as well as integrating such data and services into the Platform's user interfaces. Structured descriptive data may be generated using the stand-alone Xper² software (Ung & al. 2010; Venin & al. 2010). In parallel, several structures support the entry and storage of categorized descriptive data items. Content under categories like “Notes”, “Etymology”, “General Distribution”, etc.), as often found in Floras and monographs, can be captured as unstandardized text for each category. However, standard prescribed vocabularies may be defined for a category (e.g. IUCN status). Categories may be structured hierarchically (e.g. geographic areas), and some structures can even relate to multiple levels of a hierarchical category (e.g. occurrence status related to an individual distribution area).

Current state of the database and editorial rules

The Caryophyllales Network is still in the process of defining the editorial rules to be applied to the entire treatment of the order, so the specifications given below may change in the future (this will be indicated on the website). It is clear, however, that the full citation of sources of information as outlined above should be followed wherever possible. As the database normally provides highly atomized content, the output can be adapted to different formats with ease. The format used in the data portal or in the PDF edition ( supplemental materials (wi.48.48301_Supplement.pdf)) in many respects follows the rules formulated by Kilian & al. (2009+) for the Cichorieae Network, another initiative using the EDIT Platform. Although the Platform offers a way to enter information items such as references, author teams, person names, specimens and even scientific names as unstructured text, the eventual aim is to enter these in a fully atomized and thus re-usable form. Updated versions of the PDF-edition will be provided as new data become available, while the data on the portal always represent the actual state of the database.

References —All references cited in the treatment should eventually be entered into the EDIT Platform database in fully atomized form (i.e. with authors forming an author team assembled from persons, with articles including the title also for nomenclatural references although this does not figure in the output of these, etc.). The “in” references for parts of a work in nomenclatural citations are also a bibliographic reference. In the database, two reference records are created, one for the “in” citation of the full work with its respective Reference Type (usually a book), and one for the part of the work (type “Book section” or “Section”). For example, the nomenclatural citation “Nepenthes epiphytica A. S. Rob., Nerz & Wistuba in S. McPherson, New Nepenthes 1: 36. 2011” creates two reference records in the database. To make the publication citable, care has to be taken to enter the full last name(s) of the author(s) and the full title as well as their abbreviated standard forms.

Nomenclatural citation of taxon names (adapted from Kilian & al. 2009+) — Titles of monographs are abbreviated in conformity with Taxonomic literature, ed. 2 (TL-2; Stafleu & Cowan 1976–1988; Stafleu & Mennega 1992–2000; Dorr & Nicolson 2008–2009), but all components start with capital letters. Titles of monographs not listed in TL-2 follow the version used in IPNI, with the exception of the recently introduced inclusion of the editor in the title. Titles not mentioned in any of these sources are abbreviated as in serials. Titles of serials are abbreviated in conformity with BPH-2 (Bridson & al. 2004; online edition); titles of serials not listed in BPH-2 are abbreviated according to the standards therein defined.

Volume and fascicle numbers: volume numbers of monographs are as a rule given only if volumes have separate pagination. The same applies to fascicles of serial volumes. The publication year always refers to the actual year of publication of the page cited. If a work has been published in parts and its publication dates have been established, the publication year is that of the corresponding part and can differ from the year(s) cited on the (main) title page of that work; the latter may also be simply incorrect. In the case of known differences between the year cited in the publication and the real publication date, the former is set in quotation marks and the real value in square brackets. Series that form part of the title are normally included in the standard abbreviation for nomenclatural citations.

Authors, Persons and Teams — At present, the following rules are applied in the database: Single persons (authors, collectors, etc.) are represented by a Person record. Multiple persons (e.g. author teams) are represented by a Team record, which consists of several Team Members, which in turn are Person records. For nomenclatural authors, the standard abbreviation has to be entered in the Person record. The abbreviation must be in accordance with the authors of plant names appearing in IPNI (2004+). Full person names and initials should be entered for Persons acting as authors of bibliographic references, as collectors of specimens, and for users (i.e. those with a password for the respective database).

Specimen data — Currently, specimen data in the Nepenthaceae database are mostly restricted to nomenclatural types (see following section) and the input is not normally atomized. It is planned to link the cited specimens to digitized specimen images (e.g. JSTOR 2013+) as far as available and to atomize the data according to the aforementioned specimen data standards. Care is taken to cite resolvable and unique stable identifiers (Güntsch & al. 2017; Groom & al. 2017) of the specimens, if available.

Nomenclatural types of names — The original citation of type material from the protologue should be entered. Data on individual type specimens are gathered from different sources, such as directly from the protologues, from secondary literature, and in the revisionary process from the type specimen labels. However, currently type information is mostly added cursorily as a string in a field reserved for preliminary information. The EDIT Platform allows categorization of type specimens as holo-, syn-, neo-, lectotypes, etc., and their duplicates as iso-, isosyn-, isoneo-, isolectotypes, etc., or as original material. If no such specification is possible, they are simply indicated as “type”.

Geographic distribution and presence status of the taxon — Geographical distributions of Nepenthes species are currently entered as a text string in the field for general distribution. However, the Platform allows for the definition of geographic vocabularies and hierarchies, so the aim is to list the individual areas where the taxon is present, and to give details on the “distribution status” in that area (e.g. “native”, “introduced”, etc.). A first step could be the use of TDWG areas (Brummitt 2001) and ISO countries (ISO 2013), which are already available as standard vocabularies in the Platform. In future, ideally the distribution and geographical status will be derived from geo-referenced, individual specimen records, where present. The Platform Data Portal allows generation of area-based maps as well as dot-maps for specimen records.

Descriptive data — The descriptive morphological data obtained as a result of the parsing process of the Flora Malesiana treatments were preliminarily stored in the structure used for general descriptive data items (“factual data” in Platform terminology), i.e. as character-character state data with non-standardized states as extracted from the text. Several dozens of characters were thus defined, among them a number highly specific to the Nepenthaceae, for example 17 characters refer to aspects of the pitchers, such as, character: “Spur”, state: “7–9 mm long, filamentous, forked 4–6.5 mm from the apex”.

The full text description is stored in parallel (and shown in the portal). At present, no molecular data are included in the database.

Interaction with the World Flora Online — A close understanding of WFO necessities could be incorporated in the workflows through a close link with the Technical Working Group of the World Flora Online ( http://worldfloraonline.org) connecting the two initiatives. In a first step, the WFO technical team provided the name data (originating from TPL) including the respective WFO name IDs. WFO-IDs were then added to the names already existing in the Caryophyllales database, and additional TPL names were imported, together with those found to be missing (TPL lacks names published after 2010, most infraspecific names, and some names found in the horticultural literature). Those additional names were assigned a WFO ID from an allocation of WFO IDs given to the Caryophyllales Network. A WFO-conformant data output of the Nepenthaceae data from our database was created as a DwC-A file, which will be ingested into WFO and replace the Nepenthaceae segment of the WFO taxonomic backbone (see WFO 2017+). The next step will be the provision of the descriptive data to WFO. Acting as a WFO Taxonomic Expert Network (TEN), the Caryophyllales Network will provide regular updates to the WFO taxonomic backbone also in the future. Where a consensus of experts exists, this feedback to WFO may also be extended to provide a preferred view to descriptive data or new versions of descriptive data matching up-to-date taxon concepts. The exact procedures are still being discussed by the WFO Council and working groups.

Discussion

As stated in the introduction, the prime purpose of the Nepenthaceae treatment presented here is to demonstrate the technical and editorial possibilities provided by the EDIT Platform in the context of the Caryophyllales Network and its aims, including the publication of a “data paper” (Chavan & al. 2011). An institutional commitment for the long-term sustainability of the database and the EDIT Platform itself is provided by the BGBM. Further development of its content strongly depends on the developments in the Caryophyllales Network. Tasks that should be tackled to maintain the content include the addition of new taxa and of taxonomic changes into the database to maintain its role as the taxonomic backbone. Expert members of the Caryophyllales Network are invited to take editorship of taxonomic groups and can rely on technical assistance by the Platform team. A structured geographical scheme for the known (past or present) distribution can be added with comparatively little effort. Standardization of the terminology for descriptive characters, especially of the Nepenthes-specific pitcher characters (see Cheek & Jebb 2013; for some ecological traits, Clarke & Moran 2011) and complete coverage (where possible) of the taxa seems to be attainable when considering the published data. Transferring this into a standardized data structure and thus making it available for the generation of localized or general identification keys is a next step. Given that there is a strong community of enthusiasts for this group, assembly of images of the taxa should also be possible.

To identify gaps in our current knowledge of the taxa, as well as to inform users about possible problems in the application of the data (e.g. in the context of conservation), some kind of standardized assessments of data quality are needed. With respect to the scientific base and therefore stability of the accepted taxon concept at species level, the classification proposed by Borsch & al. (2018) may be applied by the specialists in the group.

The Plant List (TPL 2013) listed 262 plant names for Nepenthes (vs the 435 specific and infraspecific names we now treat in the database), almost all of which were considered “unresolved”. The taxonomic knowledge existing in the published literature was much further advanced, but not present in the databases that provided the base for TPL. The revision by Jebb & Cheek (1997) recognized 82 species, with 74 in the area of the Flora Malesiana. In their treatment for that Flora, Cheek & Jebb (2001) already recognized 80 species in the area, plus three naturally occurring hybrids. Of the species recognized in the Supplementary Material of this article, 32 were published between 2001 and 2010, and an additional 47 from 2011 onward. In their checklist, Clarke & al. (2018) accepted 170 species in eight sections, probably not considering four species that were published after preparation of their manuscript. The Nepenthaceae section in the Caryophyllales database includes those newly described species and accepts N. extincta and N. kurata, thus recognizing 176 species and nine naturally occurring, named hybrids.