Checklist and distribution of the species of Seychelles for conservationists

Study Extent

The Republic of Seychelles

Sampling

The species and occurrence records compiled here come from literature (taxonomic papers and monographies), grey literature (unpublished reports), specimens stored at the Seychelles National Herbarium, field work with observations and/or specimens collected individually (exploration) or together within plots (surveys, inventories). It also contains data gathered in iNaturalist within a project named "Seychelles Bio Gallery" which allows us (curators of the project) to have access to the exact coordinates of endangered species and other records obscured by the contributors. Those coordinates are obscured again in this dataset for the species that are at risk for poaching. These data are compiled in 1 taxon core table plus 4 extension tables. Below, we explain their preparation and content (use of the GBIF fields and correspondence with the BIO database). 1. Taxon core table: conceptualization, content and definitions In our BIO database, we have had a checklist of the plants of Seychelles building up over the last decade. Synonymies and taxonomic information is managed in a way very similar to the one of GBIF. The purpose of our checklist is really to provide a list as complete as possible of all taxa present in Seychelles. Therefore it includes names of taxa that are recognized by us as distinct entities but do not correspond to published names, e.g. “Vincetoxicum sp.A Senterre”. These names are indicated by the taxonomicStatus = ACCEPTED and the nomenclaturalStatus = NUDUM. In some cases, we include names that correspond to unpublished combinations, e.g. “Aphloia theiformis (Vahl) Benn. subsp. seychellensis (Clos) unpublished” or “Polyscias sechellarum Baker var. curiosae (F.Friedmann) unpublished”. The list compiled here contains only names of taxon recognized at the most detailed level, some at species level and others at subspecies or variety level. But it does not include names of upper level such as genera, which are only referred to in the column genericName. Also, when a species has several infra-specific taxa recognized, e.g. within Polyscias sechellarum, we do not include the name Polyscias sechellarum (which is vague and does not really represent an elementary biodiversity entity) but rather we list all names and synonyms related to the recognized infra-specific names. We also exclude from this list all names that have been cited in Seychelles but are and were never actually present. For these published misidentifications, we only list the misapplied name using the taxonomic writing "auct.non xx, sensu xx". Species names that are doubtful records in Seychelles but that could possibly be or have been present are kept in the list. In addition, we evaluated the sensitive nature of precise occurrence data for the species listed, and we proposed a sensitive status following the categories proposed in https://doi.org/10.15468/doc-5jp4-5g10). This list was reviewed and discussed with stakeholders in Seychelles, as well as other statuses proposed in this dataset. Finally, for the broad taxonomic classification of the listed taxa, we used the columns kingdom, phylum and class in the same meaning as the one used by GBIF. Nevertheless, when it comes to extracting information from a checklist, we find the taxonomy-oriented backbone of GBIF not so practical compared to categories such as those found in iNaturalist for example. Therefore, considering that GBIF does not have a field to accommodate custom classifications, we store our classification in the field higherClassification. Basically, our classification treats at the same level of hierarchy things like Monocotyledons, Ferns (incl. fern allies) and Marchantiophyta, rather than having Tracheophyta, Marchantiophyta, etc. (which we find too much driven by systematics and little practical for conservationists). We store 3 levels of hierarchy in higherClassification, along with a numeric code for the narrower level of the hierarchy. Therefore, the field higherClassification contains 4 columns separated by the character "|" and which we can extract in R under the names "taxon_group_id", "taxon_group", "taxon_mesogroup", and "taxon_macrogroup". The GBIF fields used or unused are as follows: -taxonID: Used to store the unique ID from the BIO database (sp_id) -acceptedNameUsageID: Used to store the taxonID of the name considered accepted in this dataset (syn_of_id) -scientificName: Used for the full scientific name including authors and taxon rank. -nameAccordingTo: Used to provide a text citation of the literature reference defining this particular use of the name (e.g. the authors of a misapplication). -higherClassification: Used to store our own classification above the family level. It contains the content of 4 columns (see introduction paragraph above). -kingdom: Used to store the Kingdom according to GBIF. -phylum: Used to store the Phylum according to GBIF. -class: Used to store the Class according to GBIF. -order: Used to store the Order according to our BIO database. -family: Used to store the Family according to our BIO database. -genus: Unused (supposed to include the authors of genus name) -genericName: Used to store the Genus name without the authors. -specificEpithet: Used to store the species epithet. -infraspecificEpithet: Used to store the infra-specific epithet. -cultivarEpithet: Unused; these have no taxonomic value and the corresponding information is rather an attribute of individual occurrences. -taxonRank: Used following GBIF standards. -vernacularName: Used temporarily for storing the preferred name, but in development with the idea to have this in an additional extension file. -taxonomicStatus: Used to designate the name representing a taxon (ACCEPTED), including names that represent a taxon not yet described or an unpublished combination. -nomenclaturalStatus: Used to describe the taxonomic validity of a name. In principle this would mostly be useful to specify some issues with names that are not accepted (some being nomenclaturally invalid). But we use it mostly to designate names that we call “morpho-species” names, e.g. “sp.A”, etc., some of which being recognized as actually taxon (which we indicate by “NUDUM”). For names (including accepted names) that correspond to unpublished combinations, we indicate “INEFFECTIVE”. -taxonRemarks: Used to present our notes on the reasons for following a given taxonomy. -informationWithheld: Used to identify species that are sensitive and require obscuring their coordinates and associated information. 2. Occurrence file The query producing this txt file compiles all occurrence data from BIO which are referring to any of the names listed in the taxon core. Below we describe our use of the available GBIF fields. -taxonID: Used to refer to the taxonID of the taxon core. -dcterms:bibliographicCitation: Used to tag all records taken from Baker (1877), Robertson (1984), Friedmann (1994) or others. But it is more or less complete only for Robertson. -basisOfRecord: Used to distinguish only between PreservedSpecimen and HumanObservation. Records that do not correspond to specimens and that have a value in the field references should take here the value "Literature". But in fact, although I found the DwC term "Literature" as correct, it is refused by the IPT; therefore I am using the term "MaterialCitation". -informationWithheld: Used to populate automatically the sensitivity as recorded for the accepted taxonID referred to (either directly or via synonymy). For sensitive records, we round coordinates to 2 decimals, eventDate to year only, coordinateUncertaintyInMeters to 1000, locationRemarks to "vague", and we obscure the fields that can be used to identify indirectly the exact location (catalogNumber, recordNumber, eventID, location, locationID). -occurrenceID: Used to store the unique ID as it is in the BIO database. This field can be used to identify an occurrence that is also part of another dataset (e.g. seyvegplot, seysensitive). For records coming from the Seychelles Bio Gallery project in iNaturalist, the ID entered here is the full path to the iNaturalist record (https://www.inaturalist.org/observations/...). -catalogNumber: Used to store a field code corresponding to the BIO field indiv_tag. -recordNumber: Used to store a specimen number in a numeric or alpha-numeric form, including "s.n." for sin numero and "u.n." for unknown numero (https://dwc.tdwg.org/terms/#dwc:catalogNumber). -recordedBy: Used for the main observer/collector, followed by associate persons, using "|" to separate each name, which are written as "Surname, Firstname" (with name particles included in the Surname, e.g. "de l'Isle, G."). -individualCount: Used to store count of individuals. -organismQuantity: Used to store a semi-quantitative evaluation of the abundance -dominance in the stand or locality. -organismQuantityType: Use to specify the use of an "abundance-dominance coefficient". -sex: Unused; maybe to search in the text of the field occurrenceRemarks. -lifeStage: Used via a translation to GBIF terms based on our stage_pop field in BIO. -reproductiveCondition: Unused although potentially through our field on phenology. -Behavior: Unused although potentially for roosting sites in some animal groups. -establishmentMeans: Used via a translation to GBIF terms based on our natural_indiv field in BIO: natural = native; semi-natural = introducedAssistedColonisation; introduced = introduced. In addition to the GBIF terms (native, nativeReintroduced, introduced, introducedAssistedColonisation, vagrant, uncertain), we distinguish two subcategories of natives: nativeEndemic, nativeEndemicPaleo. These 2 additional categories are especially used in the Species Distribution extension. -degreeOfEstablishment: Used via a translation to GBIF terms based on our stage_pop field in BIO. -pathway: Unused yet, to be translated from our own terminology. -georeferenceVerificationStatus: Unused yet, but to be developed. -occurrenceStatus: Used for presence/absence. We have also added the value "doubtful" (especially for use in the Species Distribution extension). -associatedMedia: Unused but eventually later to link images (using | if more than one). -occurrenceRemarks: Used to store the observers notes corresponding to the observation field in BIO, i.e. not related to the habitat (see fieldNotes) but to the individual, species or population observed. -Organism: Unused (organismID; organismName; organismScope; associatedOrganisms; previousIdentifications; organismRemarks). -MaterialSample: Unused (materialSampleID). Eventually, later, to store one to many duplicates and exsicata. -eventID: Available indirectly (using the occurrenceID) via the dataset seyvegplot (https://www.gbif.org/dataset/4fc42f17-eaeb-4296-949d-34b8414eb1c1). -fieldNumber: Unused here, but available in seyvegplot dataset. -eventDate: Used for the observation datetime for a particular occurrence (YYYY-MM-DD HH:MM:SS) -samplingProtocol: Used for indicating the code describing a sampling protocol in BIO (method_tag) and also available through the link to the seyvegplot dataset (see eventID). -habitat: Used to describe the habitat as provided by the source (field habitat_txt in BIO). -fieldNotes: Used for the standardized description of the ecosystem and its state. This is still in development. It includes 4 columns separated by "|": "habitatFunctional" (hab_fct), "ecosystemFamily" (hab_fa), "standNaturalness" (natural_hab), "standDegreeOfInvasion" (invasion), and the "standDevelopmentStage" (stage_hab) (see http://dx.doi.org/10.1016/j.ecocom.2021.100945). -eventRemarks: Used to store here the foreign key linking a given species observation to a given stand observation in the "ecosystemology" dataset (field georef_id in BIO). This field could also be used to store information on particular projects that led to the collection of individual occurrence records (when that data will have been cleaned), squeezed in using a pipe (|)? -locationID: Used to store the unique code for a given location at the stand scale, corresponding to the "georef" field in BIO. The unique identifier called "georef_id" in BIO is a code that corresponds to a given time slice of a stand, i.e. as described at a given moment in time. More than one georef_id can exist for a given georef, although so far we have not done such data collection on stand monitoring. The georef_id corresponds to the core identifier of stand occurrences in the dataset "ecosystemology". We include it in this dataset using the field eventRemarks (just like we did for the seyvegplot dataset). -locality: Used to describe the standardized locality in the BIO gazetteer (field "gazet" in BIO), if available. -verbatimLocality: Used to describe the location as given by the source (field "locality" in BIO). -islandGroup: Used to store the Island group as defined in BIO and which corresponds to TDWG4 (TDWG:SEY-OO and TDWG:ALD-OO), but in a modified version for the Seychelles to better reflect the Island groups as they are recognized in Seychelles. -island: Used to store the name of the island using the name considered "accepted" in the BIO gazetteer table (which manages place names synonymies). -country: Used with the value "Seychelles". -countryCode: Used with the value "SC". -verbatimElevation: Used with a value of elevation in meters, except if specified otherwise. -locationRemarks: Used to describe the coordinates precision using a text-based semi-quantitative evaluation (toCoordinate, toLocality, vague). -decimalLatitude: Used to store the decimal latitude (to be exported from Access as string from within the query to avoid a MS Access bug that round the value to the second decimal). -decimalLongitude: Used, ibidem. -geodeticDatum: Unused (GBIF will assume "WGS84", which is correct) -coordinateUncertaintyInMeters: Used (provided with smartphone data collection). -georeferencedBy: Unused, so far. -georeferencedDate: Unused, so far. -georeferenceProtocol: Unused, so far. -georeferenceRemarks: Unused, so far. -GeologicalContext: The whole section is unused. -identificationQualifier: Used to store the confidence on the current identification: "cf." (det_conf code in BIO =20), "aff." (det_conf=30), "unidentified" (det_conf=40), "identified" (det_cont=10 or ""). -identifiedBy: Unused yet, but to be used with the format "Surname, Firstname". -dateIdentified: Unused yet, but to be used with the date of determinavit. -identificationVerificationStatus: Unused, yet, to be developed. -identificationRemarks: Unused yet, but to be used with the content of the BIO field co_det!comment. 3. Species distribution extension This extension provides key metadata on species, regarding their origin status, their actual presence, in Seychelles in general, or per island group. In BIO, we do not have distribution data at the scale of island groups other than the very broad dichotomy Inner Islands vs. Outer Islands. This dichotomy has been used in many reports and taxonomic documents, although it is very little meaningful. Firstly, the so-caller Inner Islands have most often been defined including Bird and Denis islands, which do not belong to the same biogeographic element as the granitic islands. Secondly, the Outer Islands are generally defined as all the remaining islands to the south of the granitic group, although those islands are mixing at least two totally different biogeographic entities: Aldabra and the rest. It is therefore more interesting to look into infra-regional floras using island groups as we define them here, in this dataset. Nevertheless, we also want to account for the data existing in the literature regarding the Inner islands vs. Outer Islands distribution of the species of the flora of Seychelles, and the species distribution extension is being used for that. More precise distribution data will have to be derived from the occurrence table. In addition, we integrate here nuances on origin status (establismentMeans), considering that some species are native to the Inner Islands but introduced to the Outer Islands, and reversely. In that area, more thinking will be done to evaluate how to best use the occurrence core for a bottom-up evaluation of native distribution, on the one hand, and the species distribution extension for a top-down definition areas to which a species is native, on the other. At this stage, in this early version of this dataset, we put in the species distribution extension the union of all country-level metadata on the species of our checklist, plus all the presence/absence and establishmentMeans for the Inner vs. Outer Islands as approximated by the TDWG4 values TDWG:ALD-OO and TDWG:SEY-OO (here stored in the field locationID). Country-level metadata can therefore be identified by the absence of value in the field locationID. -taxonID: Used to refer to the taxonID of the taxon core. -locationID: Used to define an area (normally within Seychelles) for which a given species is given some information on its presence and its establishment mean. At the moment, we use this field to store in this extension data that we have on those attributes at the scale of the Inner Islands vs. Outer Islands as approximated by the TDWG4 values TDWG:ALD-OO and TDWG:SEY-OO (see http://rs.gbif.org/areas/). Later we might make more use of this to detail generalized metadata on species for any particular area, e.g. BIO:islandGroup:Aldabra, BIO:island:Mahé, BIO:island:Round (Mahé). -countryCode: Used with the value "SC" -lifeStage: Used based on a translation of our BIO terminology into the GBIF terms. -occurrenceStatus: Used with the values "present" or "doubtful". -establishmentMeans: Used with the GBIF terminology, with the additional term nativeEndemic and nativeEndemicPaleo. -degreeOfEstablishment: Used based on a translation of our BIO terminology into the GBIF terms. -pathway: Unused yet. -threatStatus: Used for the National Red List status, in Seychelles. -appendixCITES: Unused 4. Species profiles extension This extension is quite straightforward and derive directly from BIO with little manipulation to make it follow the GBIF architecture. It is useful for basic intrinsic information on species such as isTerrestrial, isMarine, etc. -taxonID: Used to refer to the taxonID of the taxon core. -isMarine: Used, derived from the field TB and hab_fct of BIO. -isFreshwater: Used, derived from the field TB and hab_fct of BIO. -isTerrestrial: Used, derived from the field TB and hab_fct of BIO. -isInvasive: Used, derived from the field invasiveness of BIO. -lifeForm: Used, putting together the BIO fields TB3, TB3_name, TB2 and TB1, separated by "|"; those fields representing a typology of biological types over 3 hierarchical levels. -habitat: Used based on a generalized definition of broad types of ecosystems. If a species is found typically in more than one of these ecosystem groups, the different values are separated by "|". 5. Alternative identifiers extension For all names listed (accepted ones and synonyms), prior to this GBIF publication, we have searched for the GBIF taxonKey. We used the name_backbone_checklist function in the R package rgbif. We also used the functions gbif_species_name_match (inborutils) and get_gbifid (taxize). The character string used as input was the full scientific name of the taxon with its authors’ names, including taxon rank (subsp. or var.). We then reviewed the result of the name matching function and corrected where we considered the match to be wrong. For misapplied names, we preferred to materialize these misapplications by a name reflecting that (e.g. “Tylophora laevigata auct. non Decne., sensu Robertson 1989”) thus linking that name to the actual species meant (in this example it would be the taxonID of the name “Vincetoxicum sp.A Senterre”) using the column acceptedNameUsageID. Therefore, in the same logic, we do not want to link the misapplied name to the gbif id (taxonKey) of its corresponding scientificName (in this case https://www.gbif.org/species/7649526), which corresponds to a distinct species. Similarly, we identified for each of the name listed in our dataset, the ID according to the IUCN Red List. We used and extract from the IUCN for all taxa from Seychelles with an evaluated status. If no evaluation is available under the scientific name that we consider to be the accepted one, then no IUCN ID is given for the accepted name, but only to the names (considered synonyms by us) evaluated in the IUCN. -taxonID: Used to refer to the taxonID of the taxon core. -dcterms:identifier: Used to store a given ID value (see www/gbif/gbifidnameusagekey) -dcterms:subject: Used to tell which database the ID is from (e.g. gbif or iucn).

Quality Control

The geographic coordinates and species identifications have been verified and reviewed in priority for plant species corresponding to species of high conservation value. But a lot of work remains to be done, and the database needs to be improved to better keep tracks of coordinates verification status and to better prioritise the records that most require reviewing. For most data having been collected in the field since 2008, the coordinates are correct from the source and the species identification are verified. And since early 2019, the data are being collected in the field using smartphones and a standardized survey designed with Open Foris Collect, which allows for quality control as the data are being collected (see https://www.researchgate.net/publication/353345419_BIO_KBA_3_4_Open_Foris_Collect_Survey_File_with_species_list_for_Seychelles).

Method steps

1. 1. Export from the BIO database to .txt files in GBIF format: The data published here are part of the ‘BIO’ database (see more details in the section ‘Project data’), which is still currently in the form of a MS Access file. A series of queries compile the data in a GBIF format. The BIO database main form has a button "Export data for publication on GBIF", which exports 12 queries to tables in another MS Access database file (Bio_user.accd). Five of these tables correspond to those described in the section "Sampling description" for this dataset, while the other tables are for the other datasets (which are all interconnected). Secondly, to convert these MS Access tables into txt files ready for publication on GBIF using the IPT, we used an R script and the package RODBC (see Bio_export.Rmd in https://github.com/bsenterre/seychecklist). This is because we have had issues with MS Access dialogue tool for exporting to txt file (e.g. memo and date fields). The text files are produced with the ‘.txt’ extension, fields delimited by a tabulation, text identified by "", exported as UTF-8.
2. 2. Publishing on GBIF via IPT: We log in the GBIF IPT (https://cloud.gbif.org/africa), then go to manage resources and we enter the ‘seychecklist’ resource. We delete the earlier versions of data sources, and then we load the new versions exported from MS Access at the previous step. Then we map (done automatically), each source file to the corresponding Darwin Core (incl. for the extensions). If there is any change in the metadata, we make the required changes. Finally, we ‘publish’ the new version, providing a brief text that explains the changes since previous version.
3. 3. Rerun the R script using the GBIF datasets and producing the Shiny app We run the R script that create ready-to-use files for the purpose of updating the shiny app (see GBIF_export in https://github.com/bsenterre/seychecklist). Then we update the shiny app version by re-plublishing the script (see subfolder 'app' on the github repository). This way the updated dataset is available to the users and to the Seychelles KBA NCG for further review and corrections.

At the time of the first version of this dataset, we filtered only elements of Plantae, but in future the idea is to extend this dataset to a complete checklist of all species of Seychelles, for all taxonomic groups.

1. Plantae
   rank: kingdom

The Republic of Seychelles