Home > CANBR >

archive-icon-1996Document relating to the history of the CANBR, by Bernie Hyland, ~1996

History of 'Australian Tropical Rain Forest Trees' key

Prepared for the web from an old document written by Bernie Hyland, exact date unknown.

I began my tertiary education in 1955, and completing it in 1959, and was awarded the degree of Bachelor of Science in Forestry. My tertiary education was financed by a Queensland State Scholarship and at the completion of my course I was bonded to work for the Queensland Department of Forestry for a period of five years. In January 1960 I was transferred to Atherton to work in the Silvicultural Research Section of the department. After an initial settling—in period when I participated in all aspects of the research program, I was given responsibility for the research into the natural regeneration of the prime cabinet timbers in North Queensland rain forests following logging and silvicultural treatment, which was extensively and successfully practised in those days.

A couple of years after I commenced duties in Atherton, the District Forester of the day, Mr Bill Bryant, indicated to the Silvicultural Research Officer, Mr Edgar Volck, that he thought that it would be a good idea if a manual for the identification of rain forest trees could be assembled. I was given the job but little or no direction about the format or techniques to be used. This was very unusual as research in those days was strictiy applied research with strict attention to budgetary considerations. I still retained responsibility for natural regeneration research as it was anticipated that the identification manual would be done in my spare time. I have often thought the job may have been given to me to keep me occupied and out of other peoples hair, as it was obvious to any thinking person that the job was impossible.

This project may have started in 1962, but it was not until 17-9-1963 that procedures for the accumulation of data were finalised and in place. Field sheets containing the first observations on bark and blaze features are still held at Atherton. The date on these originaisheets is 17-9-1963.

I had, early in my career, come to the conclusion that dichotomous keys were of dubious value, being difficu1t to use and of doubtful accuracy. I was convinced that the system would not work unless it was comprehensive. Although we did not have a clear idea of the number of species which would eventually be encountered, we did know that there were about 160 tree species which were utilized by the timber industry. It was therefore obvious that some hundreds of species would be involved.

I was familiar with card keys, ie edge knotched card keys, but had doubts about their suitability as it was difficult to get cards which were big enough to accommodate hundreds of species. For the job I had in mind, it was necessary for each card to have hundreds of holes around the margin. This could be done by using very large cards, but they were difficult . to use. Cards were available with a double series of holes around the margin but they were also difficult to use. I found references to Peek-a-boo or Visual punch cards in the literature, and thought they maybe suitable for the project. I knew that I would still need large cards with hundreds of punching positions, so I constructed a template out of a flat sheet of galvanised iron and drilled a large number of holes in it in groups of 100. The idea was that numerous identical sheets would be produced by the dyeline process and the appropriate holes punched manually. Photocopiers had not yet been invented, or if they had, they were not readily available hence my recourse to the dyeline technique.

I cannot remember how it happened but about this time I must have discussed the project with Neil Henry who worked in the Forestry Department head office in Brisbane. He suggested that my problems could be largely overcome by using the standard 80 column punch cards which were the standard way of assembling data for input into computers in those days. Some trial runs soon proved that Neil was right and it was decided that we should proceed along those lines. However, it was not a simple matter to punch the key characters directly onto the cards. The bark and leaf characters were recorded separately as a string of numbers, each number corresponding to a particular bark or leaf character. (We still use the same basic technique today.) Each tree species was represented by three or more individuals and each species was allocated a code number, Neil wrote the necessary Fortran programs to amalgamate all of the observations for each species and output it in a form so that the holes could be punched in the desired position in each column on the standard 80 column punch card. This was not . an easy job for the operators of the card punches, firstly because any holes punched in the wrong place could not be corrected and such cards had to be rejected and a new card substituted and punched all over again, and secondly the card punches were designed to punch no more than two holes in each column.

Despite these and other technical problems work proceeded and in 1971 the first card key was produced. It was titled “A Key to the Common Rain Forest Trees between Townsville and Cooktown Based on Leaf and Bark Features”. It consisted of a deck of cards and a book containing instructions for using the key and a collection of Spot Features to help and act as a check in the final stages of the identification process. The whole package was published by the Department of Porestry and although primarily for use within the department was also made available for sale to the public. It contained information on 584 species and included all the species which were . then considered to be of commercial value.

Neil Henry’s contribution was vital. I remember him telling me about problems they experienced duplicating the master set of cards. As mentioned previously, most machines were designed to punch no more than two holes per column. However, the master cards for some features had virtually every hole in every column punched. The duplicating punches found this job particularly traumatic and from all accounts were never the same afterwards.

Neil pointed out to me sometime in the 1960’s that it would be possible to identify specimens directly from the information in the computer database. It should be pointed out that when we started on this project there were only two computers available to us in Queensland, one was in the Treasury Department in George Street and the other was in Queensland University. They were huge machines and by current standards had vey limited capacities. To test the possibility of identification by computer, Neil got me to record the character code numbers for a few specimens in Atherton. After doing this I posted the data sheets to Brisbane where the information was punched onto cards and fed into the computer together with Neil’s program and the results, ie the names of the species, were printed out on the wide tractor fed paper which was standard in those days. This was then posted to me in Atherton to check on the correctness of the result. The whole process took about two weeks and I believe it would be true to say that we were generally considered to be quite mad. This conclusion was to some extent reinforced by Neil’s sense of humour. He once posted me two small plastic bags one contained the chads from a card punch with a label saying that they were spare chads to cover incorrecly punched holes in the cards. The other plastic bag was completely empty, but it had a neatly written label advising me that it contained spare holes for use on cards which needed additional holes.

Despite the complete impracticability of computer identification at that time, we both believed that the day would come when most people would have machines of sufficient capacity in their offices to make computer identification of plants a practical reality. If anything, the whole process happened faster than we believed possible, but we were in on the ground floor. It now seems ironical that in those days we were ahead of the technology. We had the data but the technology was not adequate. Today the technology is more than adequate but we do not have the ability (mainly staff and funding) to utilize and exploit the available technology.

In 1971 I transferred from the Queensland Forestry Department to the Forest Research Institute which had established a research station at Atherton. This was a cooperative research station between the Commonwealth Forestry and Timber Bureau and the Queensland Department of Forestry. This was a good arrangement, and the Officer in Charge, Dr Geoff Stocker agreed that, as the station botanist, I should continue work on the card key as part of my overall duties. This opened new horizons for me because I had previously been restricted to the rain forest area between Townsville and Cooktown but now I could travel up Cape York Peninsula to rain forests of Iron Range and McIlwraith Range. Although these areas are not as floristically . rich as the Towneville - Cooktown belt they do contaIn species which I had not previously encountered. To cut a long story short, work proceeded at a steady pace and in 1982 an expanded version of the key was produced. It was titled: “A Revised Card Key to the Rainforest Trees of North Queensland.” It was published by CSIRO because the Forest Research Institute was taken over by CSIRO in 1975, but it is important to state that Neil Henry was involved in this version of the key because he produced the master set of cards. This card key covered the area between Townsville and Torres Strait and included information on 799 tree species. This was virtually the limit for 80 column punch cards with ten punching positions in each column. We knew that we had reached the end of the road for punch cards and indeed industry had already started to scrap card punches. In fact we had to get the cards duplicated in the IBM museum in Boston because there was no one in Austalia willing to attempt the sob.

In 1973 I met Dr Trevor Whiffin of La Trobe University. He had decided to visit the Atherton research station after hearing about it from Professor Wardrop. Trevor was a graduate of Cambridge University but he had spent some years at Austin University in central Texas studying for his Ph.D. The thing that struck me when I first met him was his unusual accent which resulted from the combined effect of a Texas drawl and an English accent. I think we started talking about the future of the rain forest key in the late 1970’s and, after the production of the 1982 version of the card key we decided to collaborate. We discussed the matter with Neil Henry and he agreed that a different approach, using new technology, was necessary. Neil arranged for a copy of all the basic key data to be given to Trevor Whiffin and we started work on a computer version of the rain forest key. The first computer version which Trevor produced, was based on the leaf and bark data used in the 1982 version of the cardkey and ran on a Kaypro II. Kaypros were more or less portable in that they could be carried by one man and their operating system was CPM. As machines with the MS-DOS operating system became more common, Trevor converted his programs to allow them to run on the new machines which were appearing in offices all over the country.

With our seemingly unlimited capacity to accumulate and access data, we decided that we would expand the key to almost gargantuan proportions by adding flower, fruit and seedling features. When the number of species in the first computer key is multiplied by the total number of features used for each of the organs, it is soon realized that one is dealing with a huge database. In many ways this is comforting but one is constantly reminded of the opportunity to make mistakes. The coding for the features in the original card key versions was virtually all done by me, but it was obvious that I could no longer handle the expanded workload associated with coding all the featuree contemplated for the flower, fruit and seedling features. When I first discussed the matter with my technical staff, they said that they would not be able to do it because they were not professional botanists and were not familiar with all the botanical terms and morphological features. I pointed out to them that it may be difficult at the start, but I had every confidence that they would soon master the process, and when they had been doing the work for a while, they would have a better knowledge of the morphological features of the families rain forest plants than most professional botanists. Few people would now argue with the essential truth of this statement. I now find it rather amusing to hear my staff talking to visiting botanists, and explaining to them in elaborate detail, the precise morphological details which distinguish various taxa.

Initially, the data recorded in Atherton were punched and recorded in files which were copied onto floppy disks and then posted to Melbourne, where Trevor copied the data on the disks into his databases. Now the data recorded in Atherton is punched and recorded in files which are transmitted directly to Trevor at La Trobe University via the Internet. This system is fast and accurate and makes a mockery of the oft quoted Canberra bogie called “the tyranny of distance”.

Work on flower, fruit and seedling features proceeded at a hectic pace in the initial stages , and all our available resources were devoted to the task, The work was new and interesting and although we knew we had embarked on a campaign where failure could have been disastrous, we could see we were making progress. However, we evertually reached the stage where we were faced with the task of collecting missing material from rare and seldom collected species. This was a trying time, and although we seldom returned from field trips without the material we sbught, it was a stage of the project which was not particularly enjoyable. By the end of 1991, we had virtually completed the work so that we were in a position to submit it to referees prior to publication. It came as a complete surprise to us to be told by the hierachy of the division that they were not prepared to allow it to proceed to publication and that they wanted certain changes made to the system to simplify it and remove all references to botanical terms. The criticisms were obviously baseless and scarcely worthy of consideration but, since they had the power to prevent publication, I decided on a course of action which caused a few ripples in the corridors of power. I sent out a questionnaire to a wide section of the community who were familiar with and were anxiously awaiting publication of the key, asking their opinion on the criticisms which had been levelled at our work. The response was decisive and demonstrated overwhelming community support for our concept. There was an amusing facet to the potential user survey, because I not only sent the questionnaire to people in North Queensland who knew something about the subject but I also sent it to the hierachy of the division in Canberra. One day I received a phone call from a senior officer who had just received a copy of the questionnaire. and he said: “If the chief finds out about this, you’re dead.” I replied: “Well it is going to be interesting, because I have sent a copy to him.” I found out later that the officer concerned had gone to the trouble to remove the offending correspondence from the chief’s desk before he had a chance to read it, I don’t know whether the officer concerned had done me a favour or not. I soon tired of this sort of nonsense and decided to devote my energies to other aspects of the key.

This was the start of the leaf atlas. The leaf atlas was a concept which had been under consideration for some time. We thought it could be done by taking photographs of skeletonised leaves. Preliminary trials indicated that this could be done, but it was time consuming and the results rather variable. Dr David Christophel of Adelaide University was familiar with the work of Scott Wing who had X-rayed leaves, and thought it could have application in his work. David is first and foremost a palaeobotanist and I first met him when he was visiting North Queensland in 1984. David maintains that he had to make a three trips to North Queensland before he could track me down, and that he eventually succeeded only because he did not give prior notice of his intention to visit the area. David had been finding fossil leaves in various deposits in southern Australia which he believed matched species which currently grew in the rain forests of North Queensland. He was forever asking me to try to put current names on his fossil leaves. I found this process difficult, frustrating and usually inconclusive. He however only wanted me to give him a possible lead so that he could do more detailed work on the cuticle and apply a name to his fossils with more confidence.

David x-rayed all our leaf specimens and, by the end of 1992, we not only had the computer key but the manuscript of a leaf atlas to boot. The whole package was approved for publication about the end of 1992 but did not hit the streets until November 1993. It was priced at $195, which I still think is a lot of money, We sold 100? copies in Atherton before Christmas and then ordered more. So far about 750? copies have been sold all over the world, and we hope to complete the sale of the 1000 copies produced in the near future.

What does the future hold ?

With the machines and software currently available, there is virtually no limit to what we could do in this line of work, Other features could be added to the basic framework, eg David Christophel would like to add the microscopic features of the cuticle. This would permit the identification of small fragments of leaves and enable zoologists to determine the diet of rain forest animals from their droppings.

We would like to add images of various sorts to help confirm identifications. The technology already exists but we don’t have the resources to utilize it, We would like to move into image recognition. We would like to add the other life forms in the rain forest to the tree key to make a comprehensive package. We have already moved in this direction and as a result of our work at Atherton and the programming efforts of Trevor Whiffin we now have a prototype version for the rain forest shrubs.

Our databases are now so large that it is not possible to keep track of botanical name changes manually. We have overcome thi problem by writing programs which compare various databases and draw our attention only to incosistencies which we then track down and correct.

Most software is now produced with a Windows Version as an option and a big percentage of operators prefer Windows as an operating system for a variety of software applications. We plan to produce a Windows version of the tree key in the near future and Trevor has already produced a prototype version which we believe will be popular particularly with younger users.

What lessons should we learn from this project ?

The first lesson we learned was to have faith in technology. The second lesson we learned was that technological progress was still accelerating and although we were initially longing for technology which had not been manufactured, we are now in the position of not being able to fully utilize the technology which surrounds us. This is even more frustrating than the position in which we found ourselves at the start of the project. The third lesson we learned was to keep things simple and user friendly without being patronising. People are quite happy to be introduced to new terms, which are defined or illustrated, so long as they can see that they are necessary because they increase the precision of the descriptions. The computing procedures we use in our keys are designed to he simple to use and do no more than allow a wide cross section of the community to identify rain forest plants. In our opinion, this is what people want and even in taxonomic papers it would sometimes be a pleasant surprise if the dichotomous keys performed that seemingly simple task. There are various was of interrogating the databases and other workers have developed other systems. In many ways this is a not an issue. So long as the system is simple to use it will find an application and there will be ways of converting databases from one system to another. The fourh lesson learned was the value of voucher specimens. The most important aspect is the basic data on which the system is based. In our system everything is specimen based and this is vital, particularly when names change or species concepts change. It is important to remember that a species is no more than a concept, it does not exist in nature. Specimens exist but species do not. The final lesson learned was the value of luck and timing. If we were trying to launch a similar project today it is doubtful if it would be approved. The following reasons would probably be used to stop the project:

Bernie Hyland

 

^ top