A complete digitization of German herbaria is possible, sensible and should be started now

Plants, fungi and algae are important components of global biodiversity and are fundamental to all ecosystems. They are the basis for human well-being, providing food, materials and medicines. Specimens of all three groups of organisms are accommodated in herbaria, where they are commonly referred to as botanical specimens. The large number of specimens in herbaria provides an ample, permanent and continuously improving knowledge base on these organisms and an indispensable source for the analysis of the distribution of species in space and time critical for current and future research relating to global biodiversity. In order to make full use of this resource, a research infrastructure has to be built that grants comprehensive and free access to the information in herbaria and botanical collections in general. This can be achieved through digitization of the botanical objects and associated data. The botanical research community can count on a long-standing tradition of collaboration among institutions and individuals. It agreed on data standards and standard services even before the advent of computerization and information networking, an example being the Index Herbariorum as a global registry of herbaria helping towards the unique identification of specimens cited in the literature. In the spirit of this collaborative history, 51 representatives from 30 institutions advocate to start the digitization of botanical collections with the overall wall-to-wall digitization of the flat objects stored in German herbaria. Germany has 70 herbaria holding almost 23 million specimens according to a national survey carried out in 2019. 87% of these specimens are not yet digitized. Experiences from other countries like France, the Netherlands, Finland, the US and Australia show that herbaria can be comprehensively and cost-efficiently digitized in a relatively short time due to established workflows and protocols for the highthroughput digitization of flat objects. Most of the herbaria are part of a university (34), fewer belong to municipal museums (10) or state museums (8), six herbaria belong to institutions also supported by federal funds 2 Borsch T et al

The large number of specimens in herbaria provides an ample, permanent and continuously improving knowledge base on these organisms and an indispensable source for the analysis of the distribution of species in space and time critical for current and future research relating to global biodiversity. In order to make full use of this resource, a research infrastructure has to be built that grants comprehensive and free access to the information in herbaria and botanical collections in general. This can be achieved through digitization of the botanical objects and associated data.
The botanical research community can count on a long-standing tradition of collaboration among institutions and individuals. It agreed on data standards and standard services even before the advent of computerization and information networking, an example being the Index Herbariorum as a global registry of herbaria helping towards the unique identification of specimens cited in the literature.
In the spirit of this collaborative history, 51 representatives from 30 institutions advocate to start the digitization of botanical collections with the overall wall-to-wall digitization of the flat objects stored in German herbaria. Germany has 70 herbaria holding almost 23 million specimens according to a national survey carried out in 2019. 87% of these specimens are not yet digitized. Experiences from other countries like France, the Netherlands, Finland, the US and Australia show that herbaria can be comprehensively and cost-efficiently digitized in a relatively short time due to established workflows and protocols for the highthroughput digitization of flat objects.
Most of the herbaria are part of a university (34), fewer belong to municipal museums (10) or state museums (8), six herbaria belong to institutions also supported by federal funds such as Leibniz institutes, and four belong to non-governmental organizations. A common data infrastructure must therefore integrate different kinds of institutions.
Making full use of the data gained by digitization requires the set-up of a digital infrastructure for storage, archiving, content indexing and networking as well as standardized access for the scientific use of digital objects. A standards-based portfolio of technical components has already been developed and successfully tested by the Biodiversity Informatics Community over the last two decades, comprising among others access protocols, collection databases, portals, tools for semantic enrichment and annotation, international networking, storage and archiving in accordance with international standards. This was achieved through the funding by national and international programs and initiatives, which also paved the road for the German contribution to the Global Biodiversity Information Facility (GBIF).
Herbaria constitute a large part of the German botanical collections that also comprise living collections in botanical gardens and seed banks, DNA-and tissue samples, specimens preserved in fluids or on microscope slides and more. Once the herbaria are digitized, these resources can be integrated, adding to the value of the overall research infrastructure. The community has agreed on tasks that are shared between the herbaria, as the German GBIF model already successfully demonstrates.
We have compiled nine scientific use cases of immediate societal relevance for an integrated infrastructure of botanical collections. They address accelerated biodiversity discovery and research, biomonitoring and conservation planning, biodiversity modelling, the generation of trait information, automated image recognition by artificial intelligence, automated pathogen detection, contextualization by interlinking objects, enabling provenance research, as well as education, outreach and citizen science.

Introduction
Plants, fungi and algae are important components of global biodiversity, and they are the basis of all ecosystems. Specimens of all three groups of organisms are accommodated in herbaria where they are commonly referred to as botanical specimens. As primary producers of biomass, plants and algae are fundamental for ecosystem services including carbon sequestration and the provision of oxygen. Primary producers also fuel marine and terrestrial food webs supporting the existence of a huge diversity of living organisms. Fungi are the major decomposers of organic material and are obligate symbionts of land plants (mycorrhiza) and many algae (lichen symbiosis). Plants, fungi and algae are the basis for human well-being, providing food, materials and medicines. To protect and sustainably use this wealth of different organisms on our planet in line with global goals such as the sustainable development goals (United Nations 2015)*1, a knowledge base on the biosphere needs to be built that supports data-driven decision making. This knowledge should be shared with both specific users and the general public, following the FAIR principles (Findability, Accessibility, Interoperability, and Reusability, Wilkinson et al. 2016). Assessing plant, algal and fungal diversity and vegetation is a key component of biodiversity management and land use practises that maintain biodiversity. In turn, biodiversity-data driven land use planning is key to protecting biodiversity.
Information on plants, algae and fungi present in an area, as well as on global distribution patterns of species and their changes over time is of primary importance. Herbaria play a pivotal role in this respect because each collected and preserved specimen represents a verifiable record of the presence of a species in a defined place at a specific time. The large numbers of specimens in herbaria provide an ample geographic coverage and information on character variability (both morphological and molecular), and this information is continuously improving with the increasing number of specimens. Most importantly, the taxonomic identification of these specimens can be verified at any time, offering a robust knowledge base in which new research findings from the ongoing investigation of life with current methods can be continuously integrated. In this context it is important to consider that scientific approaches of elucidating the ancestral relationships of organisms and delimiting species and taxonomic groups -the field of biological systematics -has been revolutionized by the advent of (mainly molecular) phylogenetics as an evolutionary method and by electronic tools that will enable an integrative taxonomy. In addition to past and present distribution records of species, herbaria provide material samples for biological research in diverse fields (Funk 2003), including global change biology (Meineke et al. 2018, Lang et al. 2019, and are a major source of species discovery (Bebber et al. 2010). Comprehensive and free access to the information in herbaria as unique archives of biological information is thus a core task in building a research infrastructure that makes full use of the existing collection resources. This can be achieved through the digitization (i.e. the transformation of specimen information from a physical to a digital format) of the research collections that have been developed, maintained and expanded over the past 200 or more years. For further details on digitization methods and data capture see Krishtalka et al. 2016.The collaborative spirit and sense of community among botanists and botanical institutions provide a driving force for a networked and comprehensive approach of mobilizing information from herbaria throughout Germany.
The botanical research community can count on a long-standing tradition of collaboration among institutions and individuals. Herbaria are essentially forming a global infrastructure for research, with individual specialists for specific organismic groups contributing their expertise to many collaborative projects. This is fostered by the tradition of collecting and distributing duplicate specimens, and sharing of expert identifications. The botanical community came together to agree on data standards and standard services even before the advent of computerisation and information networking. For example, the Index Herbariorum*2, first published in print in 1952, serves as a global registry of herbaria, and provides a standard to uniquely identify the location of specimens cited in literature. Index Herbariorum was also among the first standards recognized by the Taxonomic Databases Working Group (TDWG) in 1986, along with early data standards covering nomenclatural literature citations and abbreviations of authors of plant names, botanical garden accession data, and an area scheme for recording plant distributions (see TDWG website*3 for an overview). TDWG evolved into the Biodiversity Information Standards organisation, now covering information for the entire biodiversity discipline.
Collaborative efforts for creating floras, i.e. inventories of plants from a defined region with detailed descriptions and identification keys, also go back a long way. A well-known example is the Flora Brasiliensis (Martius et al. 1840(Martius et al. -1906, probably one of the first truly international research endeavours undertaken. This was taken up again recently by the Flora do Brasil 2020 initiative (BFG 2018), which compiles a flora with more than 40,000 plant descriptions already published online, drawing on contributions from the global botanical community. The commitment of the global botanical community to cooperate is underlined by the World Flora Online (WFO) Initiative*4 where the world's botanical institutions joined forces to support the Global Strategy for Plant Conservation (GSPC) of the UN convention on biological diversity.
Collaboration has also been fundamental to the development of digitization in the botanical community, possibly best demonstrated by the Global Plants Initiative (GPI)*5. It started with the digitization of African type specimens in herbaria and extended to all plant types worldwide through support of the Andrew W. Mellon Foundation. Type specimens held in herbarium collections provide the anchoring point for the naming and delimiting of species. Improved access to type specimens through digitization is thus of fundamental importance in the ongoing assessment of species limits that includes the results from molecular studies. The GPI global network and digital information infrastructure is already heavily used by the global research community. Newer structures such as the DNA-Bank network (Gemeinholzer et al. 2011) and the subsequently established Global Genome Biodiversity Network GGBN; (Droege et al. 2013) have promoted and facilitated the documentation of research data connected to voucher specimens. Herbaria thus play an integral role in a modern additive research process (Henning et al. 2018) that aims to at describe and understand the evolution and diversity of organisms worldwide. In this context, digitally imaging physical specimens is a crucial pre-requisite to getting a broader user base (both the scientific community and citizen scientists) involved in curation, annotation and other research activities, recently termed Digitization 2.0 (Hedrick et al. 2019). Apart from type specimens, the data connected to all specimens in herbaria are needed in digital form for modern research. Large scale digitization of herbarium collections has therefore been implemented in many countries, e.g. in France, the Netherlands, Norway, Finland, the US and Australia (see section "Experiences and achievements in other countries"). In Germany, important steps were taken in projects financed by the Federal Ministry of Education and Research (BMBF, mainly in the context of Information Facility [GBIF]*6), and the German Research Foundation (DFG), but were largely restricted to prototyping and establishing information infrastructures, with the actual digitization efforts in part funded by the DFG LIS program. However, the actual scanning of specimens is still largely dependent on the limited resources of individual institutions. As international examples show, a concerted effort on a national scale will provide major synergies in Germany and an opportunity to create sustainable information infrastructures across institutions of different organizational setups (e.g. universities, research museums of the Leibniz Association, local public or private foundations).

Botanical Collections in Germany
Botanical collections are composed of a variety of different kinds of specimens including dried specimens mounted on herbarium sheets (higher plants and ferns) or in envelopes (bryophytes, fungi, algae), living collections of botanical gardens and seed banks, DNA and tissue collections, specimens preserved in fluids, wood, fruit, seed, fibre, and pollen collections, as well as micro-preparations on microscope slides. Worldwide as many as 388 million specimens are stored in nearly 3,100 herbaria in 178 countries. The three countries in Europe with the highest number of herbarium specimens are France, the UK, and Germany with 25.96, 22.31, and 22.16 million specimens, respectively (Thiers 2019, Table 1). Germany has 70 herbaria that are accredited according to the Index Herbariorum 2018 (global rank 4 in number of specimens per country and rank 10 in number of herbaria per country), and three of them are among the twenty largest herbaria in the world. The 22.16 million specimens listed for Germany are equivalent to 6% of the specimens worldwide and to 12% of the specimens stored in European herbaria. German herbaria are therefore not only an important research infrastructure at national level but also of global importance. In order to foster the digitization efforts in botanical collections two workshops with participants representing many of the botanical collections in Germany were held at Botanic Garden and Botanical Museum (BGBM) in Berlin in February 2017 and in February 2019. During these workshops a survey on key data on the collections in Germany was organized. Data collated and discussed included the number and kind of mounted and unmounted specimens, the available percentage and degree of digitization (e.g. ranging from label information as text in a data base to high resolution images), the state of curation (e.g. availability of staff dedicated to curation and technical management of the collection), the institutional setup of the collection (i.e. supported by a university and state funded; with partial or complete federal support, mostly in a Leibniz institute; state or municipality funded museum; or private registered association), the availability of infrastructure for digitization and data management, and the annual growth of the collections. The most recent survey among the herbaria of Germany was performed in 2019 and covered collections in 62 herbaria.

Number of Herbaria
According to this survey, the number of specimens in these 62 herbaria exceeded the number listed in Index Herbariorum for all 70 herbaria in Germany. Our survey revealed an estimated number of 22.8 million specimens that are held by these 62 institutions, of which 19.8 million specimens or 87% are not yet digitized (Fig. 1). The geographical origin of the specimens is more or less uniformly distributed over Europe, Africa, temperate Asia, North America and South America. Slightly fewer specimens stem from tropical Asia and Austral-Asia. As one would expect considering the global distribution of biodiversity and the size of the respective floras, specimens from the Pacific and the Antarctic region are less represented in absolute numbers than other regions.
Of the 22.8 million specimens held in the herbaria 53% are objects mounted on cardboard sheet (mostly seed plants, ferns, lycophytes, macroalgae), which lend themselves to imaging of the entire specimen. Another 21% are not mounted yet and 26% are stored in other forms, such as in boxes, in envelopes, on slides, or in liquids (mostly bryophytes, lichens, fungi, and algae). For these special formats photographic capture of label data is of interest, which has to be done in a separate work flow. Imaging often requires the consideration of specific objectives and may require specific expertise on the taxonomic group and special preparation to visualize characters relevant for identification. For example in the large herbarium Munich (M) 75% of the flat objects are vascular plants, and 25% are bryophytes, macroalgae, and phytopathogenic fungi. In the largest German herbarium in Berlin (B) 71% of the specimens are flat objects (seed plants and ferns), 28,5% are stored in envelopes and boxes (bryophytes, licophytes, fungi), and only 0,5% are on slides or in liquids (algae). It is reasonable to expect a similar distribution in most of the other German herbaria.
The rich academic history of German universities and the federal structure of the country today results in more distributed contributions compared to the more centralised collections within countries of similar size and history of science. The ten largest of the 62 herbaria harbour approximately 75% of the total number of specimens, whereas the majority of herbaria hold between 50 and 100 thousand specimens (Fig. 2). In 28 herbaria there are an estimated total of approximately five million specimens which are not mounted and thus Proportion of digitized specimens in the surveyed herbaria in Germany (left) and proportion of image and text data digitization vs. "text data only" (data) among the digitized specimens (right).

Figure 2.
Number of digitized and non-digitized specimens held by 62 herbaria in Germany that were surveyed in 2019 (herbarium abbreviations according to Index Herbariorum).
are not yet ready for automated processing in mass digitization efforts. This is essential prior to digitization in eight herbaria. Not all of the surveyed 62 herbaria show a growth in the number of specimens. The 31 herbaria which are increasing in size currently acquire a total of approximately 200,000 specimens per year. Approximately 79% of the yearly growth in number of specimens is achieved by the ten largest herbaria.
Most of the herbaria are part of a university (34 herbaria with a total of 14.3 million specimens). Fewer belong to municipal museums (10) or state museums (8), with a total of six million specimens. Six herbaria with a total of 2.5 million specimens belong to institutions also supported by federal funds such as Leibniz institutes, and four belong to non-governmental organisations with a total of 0.06 million specimens. A common data infrastructure must therefore integrate different kinds of institutions.

Digitization of Herbaria in Germany
Digitization within 23 herbaria in Germany has so far resulted in three million digitized specimens. These herbaria hold 19.5 million specimens, approximately 86% of the total holdings in the 62 herbaria surveyed in Germany. In most cases, only textual data (label information) have been recorded (70%). More rarely (30%), entire specimens have been photographed or scanned in addition to textual data. An average of 48,000 specimens are digitized per year; 75% are sheet-mounted specimens. Given the figure above of 200,000 accessions per year, this implies that currently only about a quarter of newly acquired specimens are being digitized (with huge differences between institutions). Only three to four service stations strategically placed in larger herbaria throughout Germany would be sufficient to digitize this annual increase in specimens in a cost-efficient way.
Most of the type specimens digitized are resulting from the aforementioned Global Plants Initiative (GPI). German collections contributed 224,324 type specimens and specimens from important historical collections to GPI. With a resolution of 600 dpi and implemented quality control for the images, as well as (mostly) detailed metadata, the GPI data are forming a high-quality resource which is made available through a common data portal. However, curation of the records is left to the individual contributors and there is no centralised quality control, especially with respect to taxonomic identification and actual type status, after the end of the project phase. Infrastructures such as the one proposed here could help to remedy that situation. Also, the establishment of globally unique identifiers for physical specimens (Groom et al. 2017 1999) is another helpful development with regard to shared quality control and data enrichment, again demonstrating the high degree of organisation and collaborative spirit within the botanical community (see also section on data capture and collection databases below).
Because of the specific distribution and setup of the botanical collections in Germany, an integrative concept for their comprehensive digitization was already envisaged in the architecture of a proposal to transform Germany's natural history collections into an integrated research infrastructure within the DCOLL project (DCOLL -German Natural Sciences Collections as an Integrated Research Infrastructure*10). DCOLL was considered a significant contribution to international joint efforts and desirable at national level by the Wissenschaftsrat (Wissenschaftsrat 2017). Unfortunately, comprehensive funding under a national roadmap for research infrastructures in Germany could not be achieved.
Despite the successful industrial-scale digitization initiatives in other countries, the use of high throughput digitization techniques like a conveyor belt or digitization station have -for financial reasons -only been an exception up to now in Germany. However, 33 herbaria (53.2%) do possess basic equipment like scanners, digital cameras, and microscopes and use them to digitize their specimens. At BGBM, a project is testing a mass throughput digitization station designed for museum objects (Fig. 3). During a period of one year, 53,171 specimens have been digitized. Two students were photographing approximately 250 sheet-mounted seed plant specimens per day and person. The workflow includes QRcoding of the specimens for automatic identification and sorting of the specimen image files and minor adjustments and mounting of the specimens. In a follow-up process basic label information like taxon names and origin of the specimens are linked to the files in the database system using the QR-Codes. Overall digitization of specimens that are not sheet-mounted still requires the development of efficient techniques and in some cases (e. g. microscopic slides) considerable curatorial effort. This is, however, especially relevant for specific herbaria that hold larger numbers of unmounted specimens. Digitizing the label information of boxes and envelopes containing bryophytes, lichens, and fungi mounted on flat card board could make use of the mass throughput workflow tested at BGBM. Other specimens would currently substantially increase the effort and time period needed and should thus rather be digitized in the longer term and driven by demand. Here, we do advocate starting the digitization of German botanical collections with the complete digitization of the flat objects stored in herbaria.

Experiences and Achievements in other Countries
The successful digitization initiatives in Naturalis (Netherlands), Digitarium (Finland), and the Muséum National d'Histoire Naturelle (France) show that major herbaria can be digitized in just a few years. The Paris Herbarium was fully digitized between 2012 and 2019 and now delivers 5,400,000 digital records to GBIF for free use by the international research community (Le Bras et al. 2017). For their size and coverage, digitization of Germany's herbaria forms an essential part of an international effort to digitize and connect natural history collections and associated data. National networks are again part of larger efforts to build a common research infrastructure. At the European level, Distributed System of Scientific Collections (DiSSCo)*11, which has been accepted on the European Strategy Forum on Research Infrastructures (ESFRI) Roadmap on research infrastructures in 2018, is an initiative of currently 119 institutions from 21 countries to turn European natural history collections into an integrated research infrastructure. Especially the European botanical collections (e.g. Meise, Berlin, Edinburgh) are currently implementing innovative concepts of semantic collection linking based on LOD-compatible identifiers, semantic enrichment and unified access to digital images. The approach overcomes the physical separation of herbaria and transforms them into a consistent information space with powerful novel access possibilities. The overall digitization of German herbaria would make an outstanding contribution to this pioneering research instrument. Taken together, this results in a rapidly developing corpus of data that provides harmonized collection information to innovative Big Data evaluation methods for textual and image information. Major achievements outside Europe are Integrated Digitized Biocollections (iDigBio)*12 in the US with more than 120 million digitally available specimen records and the Atlas of Living Australia (ALA)*13 with over 12 million specimen records.

Technical Components for Infrastructures of Scientific Collections
An overall digitization of the German herbaria requires the setup of an efficient digital infrastructure for storage, archiving, content indexing, networking as well as standardized access for the scientific use of digital objects. The digital storage and linking of botanical collection objects can be based on a standards-based portfolio of technical components already developed and tested by the Biodiversity Informatics Community over the last two decades. Therefore, the focus can be on upscaling with an extensive use of well-established components, as well as digitization protocols established at individual institutions (e.g. Haston et al. 2012). This will drastically reduce the need for developments and lead to a cost-efficient approach. Solutions for storage and archiving could be evaluated with respect to data formats and retrieval times needed for the data and may include local as well as central components. Infrastructures for data analysis with the need for high performance computing are likely to require different hardware as well as links to further data sources and can therefore be addressed in complementary and follow-up processes. The high throughput of object digitization requires effective workflows for data quality control. For this purpose, scientific personnel must be provided with the ability to check a defined number of digitized objects synchronously with the digitization process, so that in the event of quality problems, an immediate return of the batch concerned to the digitization process can be initiated.

Standards and access protocols
A digital infrastructure for the German botanical collections must be based on common data standards and access protocols with which distributed data can be uniformly queried and evaluated. Access to Biological Collection Data (ABCD, Holetschek et al. 2012) and DarwinCore (Wieczorek et al. 2012) are ratified TDWG (Biodiversity Information Standards*14) data standards for biological primary data (collection objects, observation) that achieve a high degree of unification. The new version 3.0*15 of the ABCD standard was migrated to a semantic platform as part of a DFG project*16 and now allows the formulation of "Application Schemes" with which specific requirements, such as the digitization of herbaria, can be addressed.
A number of protocols and software components have been developed and successfully implemented for access to distributed databases. These include Distributed Generic Information Retrieval (DiGIR*17), TDWG Access Protocol for Information Retrieval (TAPIR*18), the Integrated Publishing Toolkit (IPT, Robertson et al. 2014) and Biological Collection Access Service (BioCASe*19, . In addition to the ABCD community standard, the latter also supports the Lightweight Information Describing Objects (LIDO, McKenna et al. 2011) standard, which can be used to connect natural history objects to cultural history networks.

Data capture and collection databases
The overall digitization of German herbaria requires the decentralized availability of systems with which collection data can be collected, curated and connected to national and international networks. In the field of botany, various powerful software systems (e. g. Virtual Herbaria JACQ, Diversity Workbench, Symbiota, Brahms, Botalista, SeSam, 4D) are already successfully used. In recent years there has been a trend towards cooperation in system development and joint data management.
In Germany, for example, the collection management systems JACQ and Diversity Workbench are cooperative developments used by a steadily growing number of herbaria.
The JACQ system is presently used by nine institutions and covers 40% of all digitized records in German herbarium collections. JACQ is used worldwide by 41 institutions in 14 countries*20, with a total of approximately 1,400,000 records covering the entire globe geographically and constituting an important source for collection data portals such as GBIF and BioCASe.
The Diversity Workbench is presently used by ten herbaria in Germany and covers a large number of vascular plants records as well as the majority of all digitized records in German fungal and bryophyte collections (700,000 records altogether). Diversity Workbench is used by more than 30 institutions, mainly in Germany*21, with a total of approximately 16 million of botanical, mycological and zoological observation and specimen records, covering the entire globe geographically and constituting an important source for occurrence data portals such as GBIF and BioCASe.
The curation of botanical collection data in cooperative online networks has a number of advantages that are of great value for the overall digitization of German herbaria. For example, as the amount of data already entered into the systems grows, the likelihood that metadata for new objects have already been entered for the associated duplicates and exsiccatal series in other collections increases. The research-and society-relevant (meta-)data can then be jointly curated in agreed interfaces and be re-used. In addition, agreements on data entry conventions and data standards can be easily implemented in interfaces accessing various underlying database systems. Automatic and semi-automatic data quality control and semantic annotation workflows can be uniformly applied to relevant data from multiple sources. And finally, technical interfaces to national and international data networks could be developed and maintained cooperatively, which means considerable savings in costs. In comparison to other kinds of natural history collections, herbaria are particularly suitable as a pilot project for jointly operated online editing interfaces, e.g. in the context of DiSSCo, the European Open Science Cloud (EOSC) and the National Research Data Infrastructure (NFDI), due to their existing high standardization and cooperative community. Such environments for a common management of researchand society-relevant collection data will be a valuable add-on to the established in-house collection management systems with a lot of institution-specific and process-oriented functions and a high data security level.

Indexing and portals
In a predominantly decentralized data infrastructure for digital herbaria, central components must be set up that guarantee fast access to the entire data stock and provide functions that can be most effectively implemented centrally. Over the last few years, a number of portal systems have been developed for this function and are available as open source software. These include, for example, the Atlas of living Australia (ALA)*22, the NHM Data Portal (Scott et al. 2019) and the B-HIT system originally implemented by GBIF and then further developed as part of the BiNHum initiative (Kelbert et al. 2015). B-HIT allows cyclic harvesting of distributed primary data (ABCD or DarwinCore) available via the relevant access protocols (Fig. 4).
These portal systems are quality checked, cleaned, and stored in an index database for quick access. Data providers receive logs of the identified quality problems and can clean them up at source. The possibility of extending user queries to taxonomic synonyms is provided by connecting taxonomic backbone databases. The unified processing of the primary data enabled the development of standardized portals which, based on the central index, provide data access via web pages and web services. The different portals of the Global Genome Biodiversity Network*23(GGBN, Droege et al. 2013), the German Botanical Gardens Information System Gardens4Science*24 and the Index Seminum of the Botanical Garden Berlin*25, for example, use the same portal technology based on the B-HIT Index for application on an international, national and local level.

Semantics and annotation
The semantic enrichment of data with links to external resources (persons, geographic features, habitat types, exsiccatal series etc.) plays an increasingly important role in the integration of collection data across institutions. The enriched data is delivered via Linked Open Data (LOD) compatible identifiers , Groom et al. 2017 as RDF data and opened for the inference mechanisms of the Semantic Web and an interdisciplinary analysis. The development of efficient methods for semantic enrichment has started in the framework of European initiatives such as the Consortium of European Taxonomic Facilities (CETAF)*26, the EU programme SYNTHESYS*27, and the EU COST action MOBILISE*28 and is likely to develop dynamically in the coming years.
The bibliographic account of exsiccatal series "IndExs"*29 (Triebel et al. 2004) is a commonly accepted web service with information on more than 2.200 widely distributed exsiccata series (titles, editors, standard abbreviation, publication dates, references). This continuously updated standard list is ready to be included in terminology services and might be used as reference list for exsiccatae in herbarium digitization approaches. AnnoSys  was developed for the annotation of herbarium data by users via web portals and is already linked to a number of existing portals. AnnoSys can be connected to portals via a simple interface and allows registered users to add or correct standardized collection data. The annotated data are then stored on a central server and can be retrieved together with the original data. In addition, AnnoSys users can subscribe to specific topics (e. g. taxonomic groups) and receive reports on annotations that fall under these topics .
European botanical collections have created a common freely available "benchmark dataset" to evaluate and compare innovative methods of content indexing and semantic annotation (Dillen et al. 2019).

International networking
By using existing international standards of biodiversity informatics as well as established protocols and software, a digital infrastructure of the German herbaria will be compatible with international data infrastructures from the very beginning. New developments, e. g. within the framework of DiSSCo currently under development, can be rolled out uniformly in the participating systems and integrated into the existing infrastructure. The resulting services will therefore make a valuable contribution to the national and international "data space" of natural history collections. With the recently published documents under the EUproject "Innovation and consolidation for large scale digitization of natural heritage" (ICEDIG)*30, e.g. on an evolving minimum information standard on digitization (MIDS) we have a good starting point to proceed in designing data pipelines for mass digitization.

Storage and archiving
With the overall digitization of German herbaria, a common and coherent approach to the management and preservation of approximately 20 million digital herbarium images is required. For the long-term archiving of raw images, the capacities and expertise of existing institutions should be used. One option would be to archive the data in a EUDAT data center located in Germany, which offers standardized services for uploads and integrity tests. The archiving process must be decoupled from continuous digitization, so that, for example, in the event of temporary network failures or bandwidth restrictions, digitization can continue. This is achieved by setting up sufficient local buffer memory. A suitable calibration of the upload process must be carried out in cooperation with the data center on the basis of the existing technical framework conditions. In a Data Management Plan (DMP) agreed between the collections and data centers involved, it is also defined which files are generated during digitization, which files are archived permanently, and which files are temporary and can be deleted if they no longer fulfil a function in the digitization process. Fast access to the archived image data will not be required. Textual label data and any further specimen assigned (meta) data (enriched data) might be processed and archived following data pipelines as currently implemented by the GFBio collection data centers*31 (Diepenbroek et al. 2014).
A reduced file size of approx. 10 MegaByte (MB) per image corresponds to an image resolution that allows to identify most of the morphological diagnostic features used in identifying organisms to species level or assessing traits (e.g. indumentum, glands) that can be discerned in such an image. The storage of the entire digital stock of German herbaria for access would therefore mean a demand of well below 500 TeraByte (TB). Consequently, a network of 2 to 4 synchronized copies of the entire data stock, operated by the participating research institutions and provided for fast access, would be possible. Should larger file sizes be required in future years, these can be generated at any time from the long-term archived raw data. This ensures that future image resolution requirements can be met. This storage infrastructure requires the development of synchronization and load-balancing mechanisms as well as the implementation of a service interface designed to integrate the data into research workflows. Cooperation with comparable developments in the context of the DiSSCo initiative is sought.

Germany (and Beyond)
Full digital access to specimens in German herbaria and associated data in a common data infrastructure will offer a data resource that opens up research opportunities and applications in various fields of basic and applied sciences (Lang et al. 2019) as well as in academic teaching and education of the general public. The free availability of botanical biodiversity data is a prerequisite for data-driven decision making in fields relevant to achieving the Aichi Biodiversity Targets of the Convention on Biological Diversity, and the UN Sustainable Development Goals (Hobern et al. 2012). The envisaged infrastructure will provide access to objects in natural history collections and their permanently linked research data and will therefore support the knowledge generation process in biological and environmental sciences. Physical objects allow quality control (e. g. through verifying taxonomic identification). Altogether, the envisaged data infrastructure enables the implementation of the FAIR principles (Wilkinson et al. 2016).
Botanical collections in Germany have the capacity to demonstrate the utility of a completely digitized collection type by addressing urgent scientific and societal questions and by developing innovative formats for teaching, outreach and citizen participation. We showcase some examples here.

1.
Biodiversity discovery and research: A digital infrastructure of botanical collections will accelerate the discovery and description of species (Bebber et al. 2010). In the face of rapid species loss, it is necessary to accelerate and streamline taxonomic workflows. Models for a collaborative, data-based approach to tackle the taxonomic impediment do exist (e.g. , but they depend on the availability of digital specimen information and associated data and objects, such as images, sequence data, georeferences, traits and more. Taxonomy as the science of discovering, describing and documenting the diversity of organisms, elucidating their phylogenetic relationships, detecting morphological and molecular (diagnostic) characters and providing valid names, is fundamental to all subsequent research on organisms. The German Research Foundation established the Priority Program "Taxon-Omics: New Approaches for Discovering and Naming Biodiversity" (SPP 1991)*32 for a six year period. A major task in this program is the "Efficient or novel use of natural history collections through automated image analysis, genetic or genomic data from historic specimens or living collections, or new ways of comparing and quantifying traits". In the first phase of the priority program, altogether eleven projects focused on plants, fungi, lichens or algae, highlighting the importance of Botanical collections*33.
Taxonomists must be provided with cutting-edge data infrastructures and analysis tools to tackle the taxonomic impediment, and the German herbaria will make a major contribution to this goal. Through better accessibility, German collections will be used more extensively by the global scientific community, thus improving their scientific value and the level of scientific treatment.

2.
Biomonitoring and conservation planning: Digitized herbarium specimens serve as tools for quality control of biomonitoring activities (Geiger et al. 2016). Crosschecking of species identification against reliably identified specimens is key, especially if taxonomically difficult groups are concerned. Digital approaches such as virtual herbaria (e.g. Virtuelles Herbarium der Lausitz*34 or specialised portals for challenging taxa*35 ) allow for greatly enhanced accessibility and new options for data presentation. Digital collections thus constitute references for morphological and molecular reference libraries, and in connection with relevant information, such as indicator values or extinction risk, they can be used to create tools such as checklists, taxonomic monographs or factsheets for biomonitoring purposes. The use of digitized diatoms in factsheets about water quality assessment (Kasten et al. 2018) is an applied example from the world of microscopic collections, which may be worthwhile including as a special collection prototype into a digital infrastructure of botanical collections in Germany. In addition, herbarium specimens are an important source of information for documenting floristic changes (e. g. Wörz and Thiv 2015), as it can be seen in the recent checklist and red list of vascular plants in Germany (Metzing et al. 2018).

3.
Forecasts about biodiversity development: Niche modelling uses the wealth of data coming with digitization of specimens to create forecasts about biodiversity and biodiversity changes (Greve et al. 2016, Pinkernell and Beszteri 2014, de Menezes et al. 2018. This powerful tool links to the current debate about biodiversity decline. Before a species becomes extinct, it suffers a gradual, often undetected loss of its populations. Herbarium data, with their depth in space and time, are a valuable data source to create and validate respective models -if they are digitally available and comprehensively exploitable.

4.
Generation of molecular and other trait information: Extraction of DNA and generation of nucleotide sequence data is by far the most frequent invasive investigation performed on herbarium specimens. These data provide the basis for phylogenetic and phylogenomic biodiversity research (Bakker 2018) and provide new identification tools (Kress et al. 2005, Geiger et al. 2016. Digitized herbarium specimens play a dual critical role in this process: They have to be explored digitally in order to find the material to close the gaps in the molecular tree of life -and they are an indispensable basis for documentation and unambiguous identification. DNA sequence information is of limited value, unless linked, ideally in a digital fashion, to a specific specimen it has been derived from. In this context, it should also be mentioned that digital specimens are a prerequisite of linking legal documents to genetic resources and thus constitute an important component of implementation of e. g. the Nagoya Protocol (Stevens et al. 2019). Regarding molecular and chemical traits, the role of botanical living collections should also be highlighted here. By linking different botanical collection types (e. g. herbarium, living collections, BioResource Centres, DNA-banks, seed banks), an even more comprehensive, well-documented and sustainable research resource can be created. This is especially relevant in plants, algae and fungi for which the use of many "omics" methods requires living organisms which should be curated under consistent workflows with derived permanently preserved objects (i.e. a herbarium specimen linked to a plant in a living collection or seed bank). We refer for example to the national network of Botanical living collections Gardens4Science*24, which makes living collections in Botanic Gardens available as a living research resource. Likewise, some fungi can be simultaneously preserved both as a herbarised specimen and metabolically inactive culture, so that information on both phenotypic traits and genetic properties will be available for future studies (e.g. Spirin et al. 2017). Last but not least, all generated trait data, when linked back to digital specimens, enhance the data pool and allow more and more sophisticated data driven research approaches (e. g. Lauterbach et al. 2019).

5.
Using artificial intelligence for biodiversity research and collection curation: Automated pathogen detection: Plant pests and pathogens can be detected in herbarium specimens (Böllmann and Scholler 2006, Lees et al. 2011, McCain and Hennen 1986, and available digital images are an important source of knowledge. Crop diseases threat global food security. Fast and accurate pathogen detection is of high economic importance in agri-and horticulture. Digital images can be used by phytopathologists in two ways. First, reference material containing properly identifed plant pests or pathogens (viruses, bacteria, fungi, insects) creates a solid basis for deep machine learning tools. Trained on public image datasets neural networks achieved high accuracy of plant disease detection (e. g. Mohanty et al. 2016, Ramcharan et al. 2017. Second, trained on reference datasets, tools can be used to analyse retroactively the distribution of plant pathogens, their host spectrum and the diversity of symptoms. 7. Facilitating straightforward contextualisation of specimens: In the course of biographical or historical research it may be necessary to link objects of different kinds that may be stored in different collections with publications or even unpublished documents (letters, diaries). In historical times, during their expeditions, botanical explorers not only collected plant specimens but also e.g. shells, coins, artifacts, photographs, geological objects and geographical information. In an ongoing interdisciplinary project we demonstrate the added value of virtually interlinking these classes of objects*36. Obviously this will be possible only when the relevant objects are available in digitized manner. In this spirit, European natural history collections are increasingly made available to the digital infrastructure Europeana Collections*37 through the OpenUp! network*38, which was significantly driven by the Botanical community. 8.
Enabling provenance research: Exploring the cultural history of natural history collections sheds light on the circumstances under which organisms from nature have been transformed into objects through the work of scientists and thus gain their value through connected data. However, localization of specimens in geographical and historical contexts requires digital access paths independent from the taxonomic storage systems of physical objects. Provenance research has gained increasing attention in colonial contexts mainly for cultural history collections, but is also of interest for natural history collections (e. g. Rahemipour 2018, Timler and Zepernick 1987). Working towards an eye-sight collaboration between countries, the botanical community has been proactive towards balancing asymmetries between developing and industrialized nations through fostering access to collections and by committing to the fair sharing of benefits arising from the utilisation of genetic resources, as required by the convention on biological diversity. 9.
Education, outreach and citizen science: Digitizing collections increases their value for purposes of outreach and education. The Digiphyll*39 project, for example, links fossil plant species to their extant relatives -it offers an identification key for fossil leaves and makes their images digitally available. The portal is designed for students of palaeontology, but also for schools and the general public, because it provides a wealth of appealing information about the fossil plants. The portal links to the accessions of extant relatives of these fossil plants in the Botanic Garden of Hohenheim and thus attracts students and the interested public to both the digital and the real collections and it is an example for the added value of linking different collection types. The high potential of herbaria for citizen science initiatives is illustrated by the project 'Die Herbonauten'*40, where a highly committed community has formed in the service of science.

Feasibility
The rate of herbarium specimen digitization and the time needed to fully digitize Germany's non-digitized herbarium specimens is dependent on factors like digitization equipment, the material to be digitized, on digitization modalities and the availability of trained staff. The community will use synergies for retro-digitization at an industrial scale by creating highly efficient digitization centers particularly beneficial for smaller herbaria, or by making use of commercial digitization centres. Thus all herbaria can train and instruct their staff and prepare their material based on already established, efficient workflows and digitization pipelines and do not have to invest in workflow development on their own. Along with the maturity of the technical concept, clear ideas exist how to prepare for the digitization process and how to manage the new digital collection that will arise. The community has agreed on tasks that are shared between the herbaria and tasks that will remain in the responsibility of the individual collections. For example in GBIF, Germany organised an innovative decentralised model for its contributions, based on the research communities in the country. After a construction project phase, the botanical and mycological GBIF nodes are fully sustained by institutional commitments of the contributing collections. The German GBIF model was further very helpful in the harmonization of data exchange protocols which are currently further developed by GFBio. While the curation of the digital objects and the associated data as well as the responsibility for physical loans has to remain in the responsibility of the institutions (curators), the new central digital infrastructure will • Provide high-quality data and services for science and society • Centrally coordinate loans requests • Organize the continuous digitization of incoming material • Provide tools for automated data capture • Facilitate and improve curation of collections by data linkage and online annotation • Concentrate efforts for storage and archiving • Implement and improve common taxonomic reference systems • Share software development and maintenance • Implement common policies for data provision and use.
All the tools and policies that are already available or will be newly developed are in accordance and compatible with international standards, policies and initiatives like Biodiversity Information Standards (TDWG), the Global Biodiversity Information Facility (GBIF), the efforts around the Distributed System of Scientific Collections (DiSSCo) and its associated projects and the policies and developments of The European Consortium of Taxonomic Facilities (CETAF), which will ensure a seamless integration of the digital infrastructure into larger collection data infrastructures.

Conclusions
We therefore advocate starting the next, most ambitious phase of digitization of German botanical collections with the overall "wall to wall" digitization of the flat objects stored in herbaria, since 1. highly efficient industrial processes for high-throughput digitization are available and affordable due to the homogeneous form and can therefore be completed within a manageable framework, 2.
the greatest possible geographical and taxonomic coverage can be achieved with one object type due to the large number of objects stored in German collections, and 3.
herbarium collections are particularly suitable for linking further object types and the establishment of a national virtual research collection due to their central role in the research workflows of the institutions as anchors for general collection activities.
German herbarium collections are held by a highly organized community. Complete digitization can be achieved with manageable effort. The digitization process is wellestablished and the methodology is comparatively easy. The technical infrastructure is developed to a large degree and missing components can be quickly developed within confined projects and in accordance with international standards and initiatives. Mass digitization of standard herbarium specimens will provide the data to support excellent science and to satisfy urgent societal information demands. The digitization of historical and type specimens is a noteworthy example of the first comprehensive set of herbarium specimens that became available at a global level. It has been instrumental for the assessment of plant species diversity and also revolutionized the research process by providing access to specimen information across countries. Moreover, digitization of specimens for the first time opened up their contextualisation within a cultural framwork and such inter-and transdisciplinary approaches for generating novel insights are likely to grow with the development of the semantic web and mass data analyses (see our example use cases). From the beginning, the new infrastructure will contain research results of the German biodiversity informatics community on semantic indexing, cross-collection integration with Linked Open Data, semantic web annotation as well as advanced collaborative taxonomic data processing platforms. In this way, a data space is created that opens up herbaria for a variety of innovative semantics-aware applications. Most importantly, the networked initiative of managing research data in the biological disciplines (GFBio) has already established an integration of data centers (data curation and archiving) with complementary as well as cooperative software development and will certainly be at the forefront of big data analysis in biodiversity. Making a global impact, of course, requires that the data potential of German natural history collections such as herbaria is made available. The authors propose to start this initiative now in order to valorize German botanical collections as a vital part of a worldwide biodiversity data pool.