BG-BASE: Documenting the World's Biological Collections
Unlimited Power of OpenInsight Makes It Happen
If Darwin had been looking for a database application to help him catalog his plant discoveries, he surely would have picked BG-BASE.
BG-BASE (www.bg-base.com) is the creation of botanist Kerry S. Walter and horticulturist Mike O'Neal. In 1985, Walter set out on a mission: create an application that could capture all the scientific data about botanic collections of living and preserved plants, along with all the record keeping required to manage and curate those collections. O'Neal joined him four years later, and together they developed BG-BASE — a world leader and de facto standard in managing information on biological collections for botanic gardens, arboreta, herbaria, zoos, universities, conservation organizations and similar institutions.
Now used in 27 countries, noteworthy BG-BASE clients include the U.S. Bureau of Land Management, The New York Botanical Garden, Beijing Botanical Garden, the Royal Horticultural Society, the Eden Project, the United Nations Environment Program and many others.
And from the beginning, the engine that drives BG-BASE has been based on Revelation Software products.
Revelation: The Best Tool for a Big Job
"Back in 1985, when I first started to develop PC-based applications," remembers Walter, "people would ask 'Why Revelation? Why aren't you using dBase II?' Later it was, 'Why aren't you using dBase III?' And over the years it was FoxPro, and then Paradox. And now they ask 'Why don't you use Access or mySQL?" And the answer has always been the same — it's because Revelation makes the best tool to handle extremely complex data models."
O'Neal agrees. "Developing an application in OpenInsight is relatively simple compared to other database managers. Yet what we can do with it is incredibly complex. The key is Revelation's unique combination of MultiValue architecture and the use of variable-length fields. There is just no other way to create something as versatile as BG-BASE."
A Design Challenge
O'Neal now runs the BG-BASE headquarters in Maine. Walter manages their U.K. office from his base in Scotland at the world-class Royal Botanic Garden in Edinburgh, a BG-BASE site since 1990. But BG-BASE began when Walter and O'Neal were both at the prestigious Arnold Arboretum of Harvard University.
The Arnold Arboretum wanted a new database application sophisticated enough to capture every scientific detail of their world famous collection of woody plants. At the same time, the Threatened Plants Unit of the World Conservation Monitoring Centre in Cambridge, England, asked Walter if he could build a collection management system simple enough to be easily understood and used by small botanic gardens around the world.
On the face of it, that may not seem like such a big deal — but how's this for a database challenge? Plants have scientific names ranging from things like Acer to Skimmia japonica Thunb. ssp. reevesiana (Fortune) N.P. Taylor & Airy Shaw (Rogersii Group) 'Helen Goodall'. In fact, a single scientific name can be made up from as many as 38 separate fields. Often there are several scientific names for the same kind of plant, only one of which is considered correct.
Common names are even trickier. A single species can be known by 50 or more common names in multiple languages, and the same common name can be applied to different species in different parts of the world. Modern molecular techniques are challenging long-standing classification schemes for both plants and animals, so the database needs to track not only multiple scientific and common names, but must also track multiple classifications using those names.
There were many other challenges faced by Walter and O'Neal. For example, existing plants at the Royal Botanic Garden Edinburgh had been alive since 1625, so records on this massive collection (comprising 7% of all known plant species) had to show historical as well as current information. Maybe other database management systems were fine for dealing with name, address, zip and account number. But to put biological best practice into a comprehensive database application, Revelation was clearly the way to go.
Variable-Length: One Size Fits All
"When we demonstrate BG-BASE for prospective clients at institutions around the world, they're always amazed by the variable-length fields," says O'Neal. "Most of them are used to Access-type systems, where you only get so many characters to a field. If you run out of room typing into that box, you're just out of luck — you can't go any further.
"But we will open up a window in BG-BASE and start typing in a field, and typing and typing — hundreds, thousands of characters. People love that! If they have something to say about a plant or a maintenance technique or about a scientific name, they don't need weird little abbreviations so the data will fit the application, and they don't need to use clumsy memo fields. We're dealing with scientists, so accuracy and thoroughness are extremely important. That's why OpenInsight's variable-length approach is such an advantage."
"The other important thing about variable-length is that you can fully atomize your field structure while still keeping the database overhead low," adds Walter. "BG-BASE contains over 21,000 fields: 9,000 data fields and 12,000 symbolic fields spread across some 300 tables. Any particular client typically uses only about 35% of it, picking and choosing what is needed at any given time. But having it all in there keeps our development costs low because we don't have to customize for every user, and users have a system they can grow without any artificially imposed constraints. You can't afford to do that in a fixed-length environment because it would take a huge amount of disk storage to reserve the space that you anticipate you might need for a data field. But variable-length fields expand and contract as needed, and the net result is that OpenInsight only needs about 1/30th of the disk space that fixed-length systems require."
"Of course, people don't worry quite so much about storage space these days, but there's a speed advantage as well," continues Walter. "It takes time for a hard disk head to move over a large amount of blank spaces reserved for fixed-length systems. But if it's moving over densely packed data that doesn't have millions of bytes of blanks, it goes much faster. Per-record access times remain remarkably flat, no matter how large the table becomes. That has allowed us to create a completely open-ended application — and that is why BG-BASE was selected to manage information for a 25-year project to produce the only global assessment of the conservation status of all plants."
The Value of MultiValue
By using OpenInsight, BG-BASE also offers the power of MultiValue architecture. "That is so important to our users, and it's another thing that always impresses them during a demonstration," O'Neal explained. "For example, a plant can only exist in one location at a time. But over the course of the life of that plant, it may have been moved to 5 locations, or to 20 locations. In BG-BASE, we simply set that up as a MultiValue field, with the current location of that plant recorded as the first value. Likewise, many institutions conduct annual inventories to evaluate the condition of their plants. Again, these are set up as MultiValue fields, with the most recent observation located on the top line. All of this crucial information is sitting there in one record for the user to see."
21 Years And Counting!
Walter says they're sticking with OpenInsight. "Ever since we first bought a Revelation product and opened those 5-1/4" floppies, as they were then, we've never looked back. Revelation is the obvious and I think only solution for managing textual data of inherently variable nature and with very complex relationships. There's nothing that can touch the tools that Revelation has been putting out for the last 20 years."
O’Neal concurs. "We feel quite comfortable hitching our wagon to Revelation Software. They are responsive to us as developers. If we need a feature or have a suggestion, more likely than not it will end up in their next upgrade. What more could you want?”