The Journal of Botanic Garden Horticulture, No. 14
BARCODES ARE DEAD, LONG LIVE BARCODES!
IMPROVING THE INVENTORY OF LIVING PLANT COLLECTIONS
USING OPTICAL TECHNOLOGY
Reinout Havinga1 & Havard Ostgaard2
The use of barcodes for record keeping in botanic gardens has been pioneered before, but attempts
have not always been successful. It has even been claimed that, for use in living collections,
barcodes are altogether obsolete. This is dicult to imagine given the success of barcodes in
almost any professional logistic or auto-ID application. We have tried to ﬁnd the ‘sweet spots’ of
barcode use and have implemented the technology at the Hortus Botanicus Amsterdam. Integrated
with the list-making functionality in the collection management software, barcodes have proved to
be an invaluable tool in improving the quality and accuracy of the inventory.
Two problems existed at the Hortus Botanicus Amsterdam (AMD) that led to the devel-
opment of barcode-supported record keeping. Firstly, serious backlogs in the inventory
of the Garden’s plants had built up over the years. These backlogs had developed
where major or frequent changes in the composition of the Living Collection occurred.
Although there have always been well-established standards for the quality of taxonomic
information, veriﬁcations and provenance, there was no apparent strategy on how to
keep the inventory up to date. Time-consuming registration duties were frequently
overruled by other priorities. Handwritten lists of plants, if compiled at all, lingered in
the oce for years. More ecient methods and clear procedures were needed to tackle
the problem of a permanently incorrect inventory.
Secondly, and paradoxically, all plant labels in the public part of AMD were already
designated with a barcode containing the eight-digit accession number of the plant.
However, there was no barcode scanner in the organisation and there was no format
for communication with the collection database. The barcodes gave the impression of
a garden using state-of-the-art technology; however, this was purely cosmetic. This
odd situation, perpetuated for decades, ﬁnally resulted in the actual application of the
barcodes as pressure increased to take a decision about their future. It was decided that
rather than just keeping up appearances, the barcodes were, at last, going to perform the
task for which they were intended.
1. Reinout Havinga is Curator at the Hortus Botanicus Amsterdam.
Address: Plantage Middenlaan 2a, 1018DD, Amsterdam, The Netherlands.
2. Havard Ostgaard is Manager and Software Engineer at Botanical Software Ltd.
Address: 9 Priston, Bath BA2 9EB, UK.
9781910877142_sibbaldia14.indd 133 24/11/2016 13:20
134 REINOUT HAVINGA & HAVARD OSTGAARD
The second problem was attended to ﬁrst, and it seemed a straightforward one: to
enable the collection software IrisBG (Rustan & Ostgaard, 2010–2016), running on a
personal digital assistant (PDA) or desktop computer, to handle barcodes and pull up
the accession record when a code was scanned. In order to create an adequate design for
the software integration, we analysed situations in which barcode applications could be
expected to be most eective. The two key qualities of barcode technology are accuracy
and speed. A scan is much less likely to contain transcription errors than a number that
is entered manually. For example, human transcription errors in a medical research
database were measured to be between 0.01 per cent and 0.53 per cent (Khushi et al.,
2012). Speed is the other factor. A scan can instantly pull up an accession or add it to
a list of accessions. It helps to reduce the time spent on updating changes in the Living
Collection, especially when collection surveys and list making are involved.
Three elements are required to implement the technology. Firstly, the plant labels have
to feature a barcode containing the accession number. Secondly, a scanner that supports
the type of barcode chosen must be acquired. Finally, the Collection software needs to
interact with the scanner hardware. Poor integration of barcode technology with the
database software is considered to be the main hurdle in implementing this technology
(Aplin et al., 2007). Using additional software to collect and organise the scans would
challenge the speed and adoptability of the solution.
AMD had previously printed Code128 linear barcodes. Although widely adopted by
the retail industry, linear barcodes can be dicult to scan when physically damaged or
dirty. More recent two-dimensional barcodes, such as QR codes or Data Matrix (DM)
codes, implement Reed-Solomon error correction (Wikipedia, 2016), which increases
their reliability even under dicult conditions. When the contrast is sucient and the
encoded information is limited, DM codes can be printed on a very small surface and
retain good readability when using a professional scanner. They can be added incon-
spicuously and economically to the layout of nursery labels, interpretation labels or
accession tags (Fig. 1). The largest DM code used in AMD on interpretation labels
measures 6 × 14mm, while the smallest printed on the accession tag is 3 × 3mm. DM
codes also support inverse printing, which makes it easier to incorporate the barcodes
into the design of our white-on-brown interpretation labels (Fig. 1).
A 2D scanner is required for scanning DM codes. These scanners are more
expensive than classic 1D scanners, but they can scan 1D as well as 2D barcodes and
barcode orientation is of less importance with 2D scanners. After having tested a number
of dierent scanners, we found that some professional 2D scanners could deliver an
instant response from a scanning distance of approximately 30cm, whereas consumer-
grade camera and laser scanners often struggled to scan or would take seconds to process
the scan. These seemingly subtle dierences can have a big impact on the overall user
experience. Considerable system engineering was carried out to integrate the mobile
9781910877142_sibbaldia14.indd 134 24/11/2016 13:20
BARCODES ARE DEAD, LONG LIVE BARCODES! 135
collection management software with the scanner hardware. Once this had been done
we had a mobile version of the software running on a PDA that could instantly pick up
tiny DM codes and pull up the record linked to it. The mobile collection software was
redesigned to support three workﬂows using barcodes, as follows:
1. A barcode is scanned and the record is displayed so that details can be checked and
2. A barcode is scanned and only the plant status (e.g. dead, alive, etc.) can be
updated directly. Additional details are not displayed.
3. A barcode is scanned and the record is added to a list, after which the next barcode
can be scanned, enabling a list to be compiled quickly.
The third workﬂow can be used to build up a list of records in a matter of seconds. This
list can be used to update all records at once or can be transferred to the desktop software
for more advanced actions.
The desktop software was adapted to recognise scans from a wired device to ﬁnd
the corresponding accession or add the record to a list in its ‘list maker’ functionality.
Barcodes positively inﬂuenced the speed and ease of working in a number of record-
Most experience was gained in the nursery. The nursery is a very dynamic part of
AMD where plant status changes are frequent from stored, sown, germinated or pricked-
out to either established (potted or planted) or dead. It is a constant challenge to ensure
Fig. 1 (left to right) Nursery label, interpretation label and accession tag with Data Matrix codes containing
the eight-digit accession number. Photos: Reinout Havinga.
9781910877142_sibbaldia14.indd 135 24/11/2016 13:20
136 REINOUT HAVINGA & HAVARD OSTGAARD
that the recorded plant status corresponds with the situation on the ground. In 2015 we
used barcodes to record the living plants in the nursery once or twice a week (Fig. 2).
The resulting list was compared to the existing plant status and when applicable the
status was updated, for example from ‘sown’ to ‘germinated’. This yielded reliable status
accounts for the nursery with an accuracy of around three to ﬁve days. The information
was collected with very little eort and without major additional pressure on the organi-
sation. The positive impact on the reliability of the database, however, was signiﬁcant.
The second application was in mass plantings. To avoid soil exhaustion, many of the
perennials in a garden area with 18 subsections were lifted in 2015 and moved to another
subsection. The total inventory of the larger area did not need to change, but most of
the inventories of the subsections were going to be completely dierent. To update the
inventories after the reshue, a new list of plants in each subsection was composed
using the mobile software on the PDA with integrated 2D barcode scanner. The resulting
lists were used to analyse which plants were new to a subsection and which were still
in their initial location. A number of the labels in the area had not yet been provided
with a barcode. For these plants, information had to be typed in by hand, which caused
a signiﬁcant delay compared to the machine-readable labels. With the Living Collection
available online for visitors to explore, being more up to date with the inventory now
improves the online experience and enhances the proﬁle of the Garden.
Fig. 2 Stocktaking plants in the nursery using a PDA with 2D scanner. Photo: Karin de Bont.
9781910877142_sibbaldia14.indd 136 24/11/2016 13:20
BARCODES ARE DEAD, LONG LIVE BARCODES! 137
Finally, good results were achieved with barcodes in ‘on-demand labelling’ of
seasonal display plants. In 2015, we had about 40 tulip varieties on display in spring,
followed by 40 dahlia varieties in summer. To be on the safe side, a few more of these
plants were propagated in the nursery and accommodated with practical nursery labels.
When the display was opened, most, but not all, of the plants were brought to the public
area. At this point, all selected plants were listed by scanning their nursery labels. The
resulting list was then used to deﬁne which plants required new labels.
Based on the supported scenarios and the activities in which barcodes have been useful,
a number of other applications of the technology can be envisaged, such as up-to-date
registration of seasonal plantings or making periodical inventories of the total collection
(Rae, 2008). Rapid surveys of phenological events such as bud burst, ﬂowering and
fruiting (Blades et al., 2008) that help predict how plants react to climatic change
(Martin, 2014) could be carried out frequently. These kinds of registrations, which
improve the quality and value of the collection data, often lie waiting due to the costs
and eort involved. The time required to record this type of data is expected to drop
dramatically when barcode technology is applied.
Radio-frequency identiﬁcation (RFID) technology may be implemented in much the
same way as barcodes, and this technology has already been tested in botanic gardens
(Aplin et al., 2007). However, producing RFID tags is more expensive than barcode
technology and for building up experience using auto-ID solutions, the more mature
barcode technology was deemed to be more appropriate. RFID tags can easily replace or
be combined with barcodes in the future and many of the experiences from this project
are directly transferable to an RFID tag implementation.
The merits of lists
In the case of the ﬁrst workﬂow, pulling up and modifying the accession record after a
barcode scan, the increased speed of barcode scanning over typing is not relevant. In the
case of updating individual accession records, the time saved by scanning (<0.1 second)
compared to typing (2–3 seconds) is negligible as most of the time is spent viewing and
updating the data. However the third workﬂow – building up lists – is a process that is made
easier and much faster by scanning, especially as the number of listed records increases.
The workﬂow is simple: scan, go to next plant, scan, go to next plant and so on. Any other
scenario that can follow this workﬂow should be able to beneﬁt from barcode technology.
Counting plants within accessions could be an interesting scenario to explore in the future.
A key feature in the software integration is the list-making functionality. This was
pioneered in the ‘Event Management’ module of the Atlantis Botanic Garden software
9781910877142_sibbaldia14.indd 137 24/11/2016 13:20
138 REINOUT HAVINGA & HAVARD OSTGAARD
(Persoon et al., 2004) that inspired a similar functionality in version 3.1 of IrisBG (Rustan
& Ostgaard, 2012). The implementation of barcodes appears to bear fruit in situations
where the list-making ability is present in the collection software. At the Botanic Garden
of the Natural History Museum in Oslo, Norway and at Utrecht Botanic Garden, the
Netherlands, smartphones are now used to scan DM codes and compile lists of specimen
numbers in simple text ﬁles. In Oslo, the list is pasted into a ‘task’ form in IrisBG (A.
Kool, pers. comm.). In Utrecht, the list is imported into an ‘event’ form in Atlantis BG (E.
Gouda, pers. comm.). Although slightly less integrated than the solution used in AMD,
the examples from Oslo and Utrecht illustrate the importance of a list-building element
in the software, which is necessary to make the implementation of barcodes worthwhile.
We found that having machine-readable labels can speed up the inventory of acces-
sions so that it becomes a viable alternative for working with checklists based on the
existing inventory in the database (as described by Latta, 2007). The traditional approach
would be to prepare a list of plants that are expected to be present and compare this list
to the current situation. Barcodes enable a more direct approach: a list is made of plants
which are currently present in (a part of) the garden. This list is then used to update the
database inventory (Fig. 3).
In situations where eciency improves, a rebound eect can be expected: having more time
on their hands, registrars may start recording more, perhaps unnecessary, data. On the other
Fig. 3 Stocktaking workﬂows are often based on checklists reported from the database. Direct recording
of the actual situation becomes an attractive alternative when barcodes are implemented. Diagram drawn by
9781910877142_sibbaldia14.indd 138 24/11/2016 13:20
BARCODES ARE DEAD, LONG LIVE BARCODES! 139
hand, backlogs in botanic garden collection inventories are common and more recordings
may lead to a more reliable inventory list and thus improve the data quality of the collection.
Many botanic gardens tend to focus on the durability of their labels and accept
expensive materials and complicated production methods. Some gardens have adopted
a model of using labels that are fast to produce and cheap to replace, accepting compro-
mises in durability. Most gardens will probably have a mixture of both models, depending
on label type (Jebb, 2003) and institutional policy. Adequate labelling of the collection
in the ‘durability model’ is achieved by reducing the need of replacement, whereas the
‘replace model’ aims to reduce the costs of replacement. Barcode technology will aid
the latter model by supporting regular stocktaking of labels that need to be renewed.
Contrary to what Aplin et al. (2007) found, we believe that successful implementation
of barcodes is not so dependent on the durability of the plant labels.
The advantage of encoding the accession number is that the encoded information will
be stable over time. Also, accession numbers are relatively short, which allows especially
DM codes to be printed on a very small surface. Instead of mere accession numbers, it
can be attractive to include information that deﬁnes the location or the specimen within an
accession (as is done in Oslo and Utrecht Botanic Gardens). This will improve accuracy,
but it will also add complexity. Another option with the IrisBG collection software is to
print QR codes with a website URL that will guide a visitor to an online plant proﬁle. The
PDA software can use the same URL to identify the relevant accession record.
In many industries, barcode technology is used as a tool for auto-ID. In botanic
gardens, labels are often not physically attached to a plant and can become separated
for various reasons. The use of barcodes as a way to automatically identify a plant is
therefore limited to situations where labels are nailed to a tree, tied to a branch or stuck
on a seed packet. In structurally rich botanic gardens with trees, shrubberies, perennial
and annual beds, nurseries and seed storage rooms, the barcodes will support eciency
only when handled by knowledgeable horticulturists.
Barcode technology can be a very useful tool in Living Collection management and
record keeping in botanic gardens. We have demonstrated that barcodes can be success-
fully implemented when the following factors are considered.
Adoptability. Software and hardware have to be easy to use and the barcode infra-
structure should be used on a regular basis. Therefore, the designed scenarios are best
built on existing workﬂows with the possibility of entering the encoded information
manually. The aim should be to make the technology a part of everyday record keeping,
and integration with the collection management or database software is advisable.
Eciency improvements. Barcode technology has most value in scenarios where
working speed is important or data volume is high. Some record-keeping activities
that were previously considered too labour-intensive are now possible with the help of
barcode technology, especially when they involve the compilation of lists.
9781910877142_sibbaldia14.indd 139 24/11/2016 13:20
140 REINOUT HAVINGA & HAVARD OSTGAARD
Dynamic environments. Barcode technology is suited to increasing eciency in the
logistic chain and has most eect in situations where regular changes occur and need
to be registered. For quick results and a short feedback loop in the implementation
process, it is recommended to start the use of barcodes in the most dynamic parts of the
collection, such as the nursery or the annual beds.
APLIN, D., VANDERBORGHT, T., GROOM, Q., VAN DE VYVER, A., LEYMAN, V. &
EMPAIN, A. (2007). The use of barcodes beyond the supermarket: possibilities and
challenges for living collections. Proceedings of the 3rd Global Botanic Gardens Congress.
Available at www.bgci.org/ﬁles/Wuhan/PapersConserving/Aplin.pdf (accessed April 2016).
BLADES, L., HARPER, G., HELFER, S., LEE, M.L., ROGERS, L. & STEWART, S. (2008).
Rapid-survey methods for ﬂowering phenology. Sibbaldia, 6, 163–179.
JEBB, M. (2003). Cataloguing and Record Keeping for Plant Collections. PlantNetwork Meeting
at the National Botanic Gardens, Glasnevin, Dublin. Available at http://plantnetwork.org/
wordpress/wp-content/uploads/5072/records.pdf (accessed April 2016).
KHUSHI, M., CARPENTER, J., BALLEINE, R., & CLARKE, C. (2012). Development of a
Data Entry Auditing Protocol and Quality Assurance for a Tissue Bank Database. Cell and
Tissue Banking, 13, 9–13. Available at http://link.springer.com/article/10.1007%2Fs10561-
011-9240-x (accessed April 2016).
LATTA, J. (2007). The use of hand held tablet laptops to record living collections. Sibbaldia, 5,
LEADLAY, E. & GREENE, J. (eds) (1998). The Darwin Technical Manual for Botanic Gardens.
Botanic Gardens Conservation International (BGCI), London.
MARTIN, S. (2014). Climate ready? Exploring the impacts and lessons from recent extreme
events at Royal Botanic Garden Edinburgh for climate change adaptation in the horticultural
sector. Sibbaldia, 12, 155–170.
PERSOON, J.G.M., GOUDA, E.J. & OTTER, J. (2004). Atlantis-BG a standard for
comprehensive, web-based collection management. Proceedings of the 2nd Global Botanic
Gardens Congress. Available at www.bgci.org/barcelona04/abstracts/pdf_abstracts/persoon.
pdf (accessed April 2016).
RAE, D. (2008). The value of living collection catalogues and catalogues produced from the
Royal Botanic Garden Edinburgh. Sibbaldia 6, 115–131.
RUSTAN, Ø. & OSTGAARD, H. (2010–2016). IrisBG – Botanical Garden Collection
Management, version 3 (software). Botanical Software Ltd, Bath. Available at
RUSTAN, Ø. & OSTGAARD, H. (2012). IrisBG – Botanical Garden Collection Management,
version 22.214.171.12411 (software). Botanical Software Ltd, Bath.
WIKIPEDIA (2016). Reed-Solomon error correction. Available at https://en.wikipedia.org/wiki/
Reed%E2%80%93Solomon_error_correction (accessed April 2016).
9781910877142_sibbaldia14.indd 140 24/11/2016 13:20
The Journal of Botanic Garden Horticulture, No. 14
POSTCARDS FROM THE FIELD: THE ROLE OF PARTNERSHIP
AND HORTICULTURE IN PLANT CONSERVATION IN SOUTH-
EASTERN UNITED STATES OF AMERICA
Jennifer M. Cruse-Sanders1
Increasingly, botanic gardens and arboreta are highlighted as eective partners to conserve plant
species diversity and restore natural communities at a time when the need for these activities
has become more urgent. Capacity for restoration and conservation at botanic gardens comes
directly from sta expertise for horticulture and research. Botanic gardens make good partners for
connecting botanical science with conservation practice. They are in a position to communicate
information about rare plant species to owners and managers of public and private lands, and
they can be instrumental in creating networks for eective conservation action. Several examples
from south-eastern United States of America illustrate how this has been put into practice. These
examples provide evidence that eorts to expand collaboration between federal agencies, states
and non-governmental organisations can lead to eective alliances to conserve plant biodiversity,
especially when plants receive a disproportionately low share of resources for conservation.
Increasingly, botanic gardens and arboreta are highlighted as eective partners for
conserving plant species diversity and restoring natural communities (Hardwick et al.,
2011; Shaw et al., 2015). The need for these activities will only become more urgent in
coming years as one in ﬁve plants face extinction and we realise the impacts of global
climate change (RBG Kew, 2016; Thuiller, 2007). Capacity for restoration and conser-
vation at botanic gardens comes directly from sta expertise in horticulture and research,
including experience in seed storage techniques, propagation and insight for identifying
appropriate microsites for augmentation and introduction into natural habitats. This is
particularly helpful because re-introducing rare plants into the landscape requires under-
standing of a species’ biology and ecology (Falk et al., 1996). Furthermore, gardens
have resources for maintaining georeferenced databases and researching taxonomic and
population genetic diversity, as well as programmes for training and outreach. In eect,
gardens are ready-made centres for conservation as places with botanical and horticul-
tural expertise at a time when we need them more than ever.
The Global Strategy for Plant Conservation (GSPC) outlines targets for plant conser-
vation to be reached by 2020 (CBD, 2012). Several of the recommended approaches to
1. Jennifer M. Cruse-Sanders PhD is Vice President for Science & Conservation at Atlanta Botanical Garden.
Address: 1345 Piedmont Ave NE, Atlanta, GA 30309, USA.
9781910877142_sibbaldia14.indd 141 24/11/2016 13:20