"Flora of Moscow" Information System on Moscow Digital Herbarium web-platform (research project #19-34-70018).

The MHA Herbarium holds vast collections from the Moscow metropolitan area (City of Moscow and Moscow Oblast) collected in the last 70 years, whereas the Moscow University Herbarium (MW) holdings are fairly evenly distributed in time over a period of 200 years. Altogether, MW and MHA have 130,000+ specimens from the Moscow Region, which make it the most densely-sampled territory across Russia. The idea of the research proposal was to digitise and precisely georeference this large dataset for proper understanding of changes in the flora around the City of Moscow through time and space. From March to October 2019, the MHA Herbarium team imaged 49.7K herbarium specimens of vascular plants at 600 dpi using Microtek 1600 Object Scan. In the next few months, 78.6% of them were georeferenced.

Further, the MHA Herbarium published online 15K images of its Eastern European collections (imaged earlier in 2017-2018), which are especially strong in the semi-arid flora of the Lower Volga Region. The MHA Herbarium collections are fully available in GBIF, Moscow Digital Herbarium and newly-established "Flora of Moscow" website (https://moscow.depo.msu.ru). At the moment, MHA Herbarium is the second largest imaged herbarium of Russia.

To schedule and perform the digitisation of the MHA Herbarium, we used five key stages by Nelson et al. (2012):

• pre-digitisation curation and staging,
• specimen image capture,
• specimen image processing,
• electronic data capture,
• georeferencing specimen data.

1. Pre-digitisation curation and staging

The section curator reviews all incoming physical accessions for meeting the basic requirements of the herbarium specimen. A specimen should be a high-quality dried plant (or several individuals) with a label bearing identification, collection site, habitat, collection date and collector. After that, unmounted new material is frozen at a temperature of –30°C for 14 days as a quarantine procedure against specific herbarium pests and then mounted. New collections are counted (and listed in the collection journal) right after mounting. Sorting and incorporation of new material takes place once a year, usually in the autumn-winter period. Right before imaging, pre-ordered self-adhesive barcodes with an acronym and a seven-digit number (e.g. MHA 0 002 094) were attached to the herbarium sheet.

Eastern European section. In December 2017, with the purchase of a specialised scanner Microtek ObjectScan 1600, we began the imaging of vascular plants in the MHA Herbarium. Since the specimens from European Russia and adjacent states constitute the largest and most used section of the Herbarium, we decided to start imaging from this section. If there were two or more taxa on a single sheet, they were remounted on separate sheets.

Moscow section. From March to October 2019, the Moscow section of the Herbarium was imaged in line with the work under the RFBR grant "Information system Flora of Moscow on the platform of the Moscow Digital Herbarium" (under A.P. Seregin). The imaging of the Eastern European section was suspended for this time. Within the framework of the project, N.M. Reshetnikova (MHA Herbarium) and S.R. Mayorov (Moscow University) thoroughly revised the taxonomy of all Moscow specimens. After that, we checked the nomenclature on the folders and corrected it against the accepted backbone.

2. Specimen image capture

Specimens were imaged in accordance with international standards with a resolution of 600 dpi and a colour checker (24 colours). After scanning, each image was automatically renamed according to the barcode served as an unique identifier. In total, 14,274 specimens of the Eastern European section were digitised in 2017–2018. Imaged Eastern European collections at that time were stored on external discs without online access.

The Moscow section was scanned more intensively under the time limit from March to October 2019. Every day, two to three operators worked on the single scanner in shifts. For each shift lasting four to five hours, 140–160 specimens were digitised. Thus, 300–400 specimens were imaged per day per scanner. In total, the herbarium team imaged 49,621 specimens within eight months and completed the mission.

During the imaging, we encountered a number of minor issues:

• Some specimens have large plants covering partly or fully the label text. The specimens were imaged as they are, whereas the labels will be captured not from the image, but from the physical specimen later.
• Sometimes two different species were mounted on a single sheet. In such cases, if possible, the specimens were remounted on to two sheets. If the remounting was impossible or impractical, the single sheet was scanned, but the image was duplicated and each file was assigned an additional digit ("-1" or "-2") to facilitate unique identifiers for each species.
• Labels of a larger size widely used in the exsiccates "Herbarium of the flora of the USSR" were often folded during mounting. We tried to remount such labels to make text fully available on images, but in some cases, the label partly covered the plant.
• In some cases, two or more parts of the same large plant were mounted on several sheets bearing a single label and further notes like “sheet #2”, “sheet #3” etc. These sheets were initially inserted into the cupboards after being fastened with a removable paper clip. However, they have been mixed over time with other specimens, so now it is impossible to trace the correct label for these multiple “sheets #2”.

3. Specimen image processing

While scanning, the operator started a new directory for every species and named it against a folder name. Before uploading the images into the Moscow Digital Herbarium, the structure of the directories was converted into a table of metadata. Thus, for each accession, the initial metadata included ID (barcode identifier), taxon name from folder without taxonomic authors and the geographic code of the area.

The taxon name, according to the protocols of the Moscow Digital Herbarium, was automatically matched with the latest version of the Catalogue of Life (CoL), from which the complete accepted name, synonymy and hierarchical list of supraspecific taxa were downloaded for every entry.

Publication of images with brief metadata is a powerful tool for rapid online access to the scanned herbarium collections. This approach was largely used in Paris where the largest herbarium of the world was imaged and published online (Le Bras et al. 2017). Similar protocols were adopted in Edinburgh and Moscow University (Haston et al. 2012, Seregin 2016, Seregin 2018).

After online publication of the Moscow Region specimens in the Moscow Digital Herbarium and GBIF, other sections of the MHA Herbarium will undergo the same procedure. Thus, to date, 64008 images of specimens of vascular plants from the MHA Herbarium are available online.

4. Electronic data capture

After online publication of the images and associated brief metadata, we link the records with existing full-label data capture of 7,087 specimens of the Moscow section (14.3%) made earlier by T.G. Nosova and I.A. Kravtsov in the form of a Microsoft Excel spreadsheet. For the remaining 42,534 specimens, the operators of the Moscow Digital Herbarium entered mandatory metadata—the collection date, the first collector, curatorial area and coordinates (if present on the label).

Similarly, a table with full-label data capture for 11,716 specimens from Eastern Europe (82% of the scanned ones) made by E.A. Karakina and B.L. Oshovskaya was uploaded as well. For the remaining 2,572 specimens, the operators of the Moscow Digital Herbarium and employees of the Garden entered additional mandatory metadata.

Thus, the minimum obligatory set of metadata available for all digitised specimens of the MHA Herbarium in this dataset include barcode ID, complete taxonomic information, collection date, the first collector, curatorial area and geographical coordinates (if available on the label). Additionally, 18,803 specimens had full-text inscriptions of labels (29.4%) due to earlier efforts.

Further full-text data capture was carried out by the operators of the Moscow Digital Herbarium for specimens collected within the City of Moscow (15,982 specimens). An operator entered the label data from the scanned image into an Microsoft Excel spreadsheet with 30 standard fields (including some pre-filled ones to avoid mistakes). Additionally, a commercial partner under the GBIF contract (2019) made full-text transcriptions of 4,617 specimens from the City of Moscow and Moscow Oblast.

After data entry, the scientific supervisor of the Moscow Digital Herbarium checked the spreadsheets for technical issues by a set of automatic, semi-automatic and manual operations. The IT-team, using the data migrator programme, then converted data from the Excel spreadsheet to the PostgreSQL database of the Moscow Digital Herbarium for further data storing and retrieving. This stage also includes some automatic checks of data consistency.

As of May 2020, the full text of labels has been entered for 39,448 specimens of the MHA Herbarium (61.6% of the imaged ones)—27,783 specimens of the Moscow section and 11,665 specimens of the Eastern European section. Full-text label transcriptions help to optimise the further georeferencing by combining labels with identical text into groups.

5. Georeferencing specimen data

The operators of the Moscow Digital Herbarium and the Garden employees carried out manual georeferencing with further implementation of the ISTRA system (Intellectual System of Toponymic Reading and Attribution), several lines of the code being written in JAVA. This code is integrated into the Moscow Digital Herbarium and unavailable as a stand-alone product.

The first algorithm of the ISTRA system combines the specimens into the groups according to the matching of the captured label text. In this case, there are two options for combining—complete matching mode and letters-only mode. The results do not differ in accuracy from the manual georeferencing. The second algorithm of the ISTRA system forms the specimen groups according to the matching of three fields—collection date, collector’s surname and curatorial area. Within the walking-day route, the standard georeferencing accuracy in most cases does not exceed 5 km. Further data refinement will help us to replace automatic georeferencing with the more accurate manual one.

In both cases, the operator inserts the coordinates manually and the system sets the coordinates automatically for all specimens of the group. The first algorithm takes precedence over the second one. In all cases, we save the log file and note the georeferencing method in the form of the standard disclaimers:

• captured from the label;
• set manually by the operator;
• set automatically by matching of the label text;
• set automatically by matching of the collection date and collector.

Manual georeferencing is carried out using standard e-cartographic libraries (Yandex.Maps, Google Maps, Wikimapia, SAS.Planet etc.) for modern specimens, whereas historic collections are georeferenced using the libraries of scanned maps (etomesto.ru, loadmap.net etc.) following the principle “collection date = map date”. Coordinate precision (rounded to 100 m) is set and stored for each manual georeferenced point.

Complete georeferencing of the specimens from the City of Moscow was a key task in 2020 for the Moscow University team (according to the Moscow project), whereas employees of the MHA Herbarium georeferenced specimens from Moscow Oblast and Eastern Europe (starting with the most prolific collectors). In total, 50,324 specimen have been georeferenced (74%), including 49,732 specimens from Russia.

For 7,414 specimens, the coordinates were taken from the label (14.7% of the number of georeferenced ones), for 10,849 specimens (21.6%), they were set manually and for 32,061 specimens (63.7%), they were calculated automatically using the ISTRA system.

The Eastern European section of the MHA Herbarium has its focus on European Russia (Table 2). The most sampled areas are the Moscow Region, Lower Volga, Central and Central Forest-steppe Regions (Table 3), the areas intensively studied by the Garden staff. Initially, these Regions were inextricably linked with the activities of A.K. Skvortsov, whose fruitful initial collections often formed a solid basis for the later extensive floristic research.

Moscow Region forms its own section in the MHA Herbarium. This is due to the location of the Garden in the City of Moscow. One of the initial missions of the Herbarium was precise documentation of the local flora, including long-term observations of both native and alien plants. Based on these materials, standard flora and checklists were published by the Garden staff in collaboration with Moscow University (Voroshilov et al. 1966, Ignatov et al. 1990, Mayorov et al. 2012, Mayorov et al. 2020).

The Lower Volga Region was one of the focus areas for A.K. Skvortsov, his graduate students and the Herbarium employees. This activity resulted in the published volumes of the “Flora of the Lower Volga” (Skvortsov 2006, Reshetnikova 2018). The Central Region is also well-represented in the collection due to another long-term floristic interest of A.K. Skvortsov, which was later continued by the Herbarium staff who published “Kaluga Flora” (Reshentnikova et al. 2010). Belgorod Oblast in the Central Forest-steppe Region is a new area of the research headed by N.M. Reshetnikova. Other Regions are less represented and are not as complete and thorough. Usually, these are either collections from various field trips of the Garden staff or gifts.

Geographical coordinates for the dataset frame are given below.

44.5 and 77 Latitude; 19.5 and 69.5 Longitude.

The dataset covers vascular plants of Eastern Europe, both native and alien species. There are also some specimens of cultivated plants, especially from the Moscow Region. The Moscow section is completely digitised and can provide figures on the taxonomic representation of the MHA Herbarium collections, whereas 14% of the the Eastern European section has been digitised and, therefore, information on its taxonomic composition only shows which families have been digitised so far.

Until 2017, the taxonomic backbone of the MHA Herbarium was the standard checklist by Czerepanov (1995). In some cases, the section curators could deviate from this source depending on their taxonomic expertise. V.D. Bochkin in the Moscow section and N.M. Reshetnikova in the Eastern European section adjusted the deviations. In 2019, with the preparation of the Moscow collections for imaging, we revised the names on folders against The Plant List (TPL), superseded shortly by Plants of the World Online (POWO). A similar approach was subsequently used before imaging of the Eastern European holdings with some deviations emerging from either the standard regional flora or new taxonomic monographs (Pimenov and Ostroumova 2012, Tzvelev and Probatova 2019, etc.).

Moscow section. It includes 2,261 species and hybrids from 773 genera and 138 families. The flora of the Moscow Region (the City of Moscow and Moscow Oblast) currently enumerates 2,363 taxa of vascular plants, including both wild and unintentionally-introduced plants (Shcherbakov and Lyubeznova 2018). We could explain a high diversity of the Moscow collections by the incorporation of purely-cultivated plants practised by the researchers of the alien flora (in case these species were found outside of cultivation). The section also includes some reference specimens of the Main Botanical Garden exposition. For taxonomic coverage across the leading taxa, see Table 4, Table 5 and Table 6

Eastern European section. In this section, 14% of the collections have been digitised so far, therefore Table 7, Table 8 and Table 9 show not the taxonomic diversity of the section, but an overview of imaged specimens. The specimens were scanned one by one following the order of the physical collection—pteridophytes, gymnosperms and angiosperms following Engler’s system against the standard catalogue (de Dalla Torre and Harms 1900). Thus, Pteridophyta, Equisetophyta, Lycopodiophyta, Pinophyta and monocots from Typha to Scirpus radicans (Dalla Torre's numbers from 1 to 468) were imaged. In addition, the genus Crataegus (Rosaceae) was digitised out of sequence.

Other

This work is licensed under a Creative Commons Attribution (CC-BY) 4.0 Licence (http://creativecommons.org/licenses/by/4.0/). The licence covers images of the herbarium specimens deposited in https://plant.depo.msu.ru/ and available in GBIF, as well as their metadata.

MHA Herbarium: Eastern European collections of vascular plants

https://doi.org/10.15468/827lk2

https://www.gbif.org/dataset/af5f680a-e0cc-46c8-b623-ceeaab70aa9e, https://depo.msu.ru/ipt/resource?r=moscowmha

1