Assessing Current Australian Python Taxonomy Using Mitochondrial Cytochrome b.
The family Booidea contains three subfamilies including the subfamily Pythonidae, the true pythons (Boidae and Pythoninae according to Barker1994). The Pythonidae is an old world group of ancestral constricting snakes (http://www.weboflife.com). There are currently 26 living Python species recognized across the globe. Fifteen species are found within Australia’s boundaries and 10 species are found only within it’s boundaries. Due to the enormous size of the Australian Continent, and the sparse human population over much of this range, many species live in geographic isolation from humans. This physical isolation as well as legislative isolation due to strict faunal protection laws, which have severely stifled herpetofaunal research, has led to confusion in Pythonidae taxonomy. The classification of pythons is still under debate. The literature rarely agrees as to which snake belongs where and even less often can agree on the scientific name for the snake in question. The past taxonomic studies were based on morphological characteristics such as scalation of head and body, teeth placement, skeletal structure, immunological evidence, and several other internal/external morphological and behavioral characteristics. At this time, I could not find any phylogenies based on molecular evidence.
A.G. Kluge at the Australian Museum did one of the most current and widely accepted taxonomies in 1993. He compared 121 behavioral, internal, and external morphological characteristics of 24 out of 26 species of pythons. Due to the size and completeness of his systematic analyses, as well as its current date, these phylogenetic relationships have become reasonably well accepted. Because of this, Kluge’s analysis is used as the comparison to the phylogeny based on mitochondrial cytochrome b as constructed for this paper. The genetic phylogeny has huge potential to extend the accuracy of current phylogenies of all organisms. Just as the current phylogenies are based on visible differences, the molecular analysis compares the differences between organisms below the level of visible characteristics. Although DNA shapes the organism’s structure, there are often nucleotide changes that do not show up in the organism’s phenotype. This can be easily demonstrated in heterozygous animals whose genotype differs from phenotype. One of the reasons for this is the fact that in many cases, the third base of the codon triplet doesn’t change which amino acid is coded for. Because the phenotype has not changed, this mutation would go unnoticed under morphological or even protein analysis. This is important in finding the ancestry of seemingly identical populations of closely related species. The effect of this “wobble” is even more pronounced as the numbers of these “silent mutations” increase. Because DNA analysis is currently the only way to find these differences, it is theoretically superior to the current methods.
Method of DNA Analysis:
The nucleotide sequences of the snakes were obtained from the Gene Bank, the National Institute of Health genetic sequence database. For the first time around, I entered the species into MegAlign by genus. MegAlign is a program which collects the nucleotide sequence off the NCBI Gene Bank website and allows the user to run statistical alignments of the sequences. This program shows the similarities of the sequences. The Gene Bank had 35 sequences for python cytochrome b from 17 different species. Cytochrome b was by far the most prevalent sequence data in the pythonidae, and therefore it was the sequence that was used for each species. There were 35 individual records, which were placed into groups comprised of 6 genus (Aspidites, Apodora, Morelia, Leiopython, Liasis, Python). Two currently recognized genus, Antaresia and Bothrochilus, were not represented in the Gene Bank. The MegAlign software aligned the sequences and then PAUP was used to calculate the best tree.
The first few trees were calculated using a fast running method of parsimony. The results gathered from the first few tests were skewed according to anyone else’s phylogenies and according to common sense. Some of the species which had multiple listings were on totally separate branches of the tree. When a more accurate method of analysis was tried, maximum likelihood analysis, the computer would have taken a few days to complete the program. (I left it running overnight for 10 hours but the next morning, it was not even finished with the first tree out of 100.) To try and get more clear results in the time I had available to me, without using a computer for days in the computer lab where another student may need access during a class, I decided to narrow my field of study. First, the species with multiple listings were looked at in MegAlign to see the degree of similarity. If there was relatively little difference, I just kept the sequence that was the longest in order to preserve as much information as possible while deleting the other sequences within the same species. All possible sequences were retained for two groups of snake, the Leiopython, and Apodora genus, which used to be included within the Liasis genus. The multiple sequences were thought to be able to more accurately describe whether or not there was justification to separate these two species into distinct genus. The 21 remaining sequences were run through PAUP using parsimony analysis. The results gathered from this test were again unsatisfactory, so another overnight maximum likelihood analysis was performed. After this 12.5-hour run, the results were noticeably less cluttered, and reasonably acceptable. To further lessen the numbers and also to concentrate on the evolution within the Australian species, the sequences for Python regus and Python reticulatus were removed from the study because their sequences were only 300 nucleotides long while nearly everything else was either 714 or 1114 nucleotides in length. The process was again completed and the analysis ran overnight this time taking less than 9 hours (yippee).
To determine bootstrap values, which take roughly 100 times more time than the method used to create the tree, I was limited again by time. There was no possible way to wait 900 hours to allow the computer to assess bootstrap values for the maximum likelihood HKY85 analysis that was previously performed; so another analysis method must be used. A neighbor joining analysis was performed and the bootstrap values were calculated using this tree. The problem with this was that the maximum likelihood tree and the neighbor-joining tree did not align very well, and therefore the bootstrap values, which determine the amount of certainty for each clade, were not matched to my most accurate tree. I have included the bootstrapped values but not attempted to relate them to the maximum likelihood tree.
Figure 1. The phylogeny of Pythonidae according to Kluge, 1993.
Figure 2. The phylogeny as calculated by PAUP, using maximum likelihood HKY85.
Figure 3. The bootstrap values applied to a neighbor-joining analysis with the same criteria as in Figure 2.
The molecular phylogeny created for this project mainly achieved its goal by coming somewhat close to the phylogeny suggested by Kluge 1993. The interesting places however, occur where they differ. In the DNA analysis, Aspidites was used as the outgroup because it is nearly universally accepted as the most ancient genus within pythonidae. This was concluded by past morphological data that showed all python species, except the genus Aspidites, as having heat sensitive thermoreceptors used to home in on prey. By using this widely accepted ancient genus, the rest of the phylogeny would not be altered by using an outgroup from the Python genus as I was first trying.
The DNA phylogeny has the Leiopython (Liasis) albertsii as a distinct clade, with no other genus on the same branch. This reinforces the view that Kluge had about the need to develop a new genus separated from Liasis. The Apodora (Liasis) paupuana on the other hand was in a clade with other Liasis, namely Liasis olivacea, and therefore may not be a distinct genus separate from Liasis. The placement of two of the python species, Python sebae and Python molurus was concurrent with Kluge’s idea that the Python genera are the furthest removed from the Aspidites. The species Python timoriensis was in the middle of a large clade of Liasis, suggesting that it may be misnamed, directly descended from a Liasis species or falsely entered in the NCBI database. This is similar to results postulated earlier that the Python timoriensis was actually an intermediate between Python reticulatus and Morelia (Python) amethistina (http://www.smuggled.com/auspyt4.htm).
There was no real separation between the Morelia and the Liasis in my DNA phylogeny. Because of this, this paper will make no real assertions about what goes where. It appears that Morelia amethistina and Morelia boeleni are alone in their own clade. This is partially concurrent with a 1975 paper by S.B. McDowall, which suggested that Morelia (Liasis) boeleni was an intermediate between Morelia (Python) amethistina and M. spilota. The tie to M. spilota doesn’t occur on my phylogeny, but Morelia amethistina and Morelia boeleni are seemingly together and distinct from other species.
Although the validity of my data is somewhat questionable due to my willingness to accept Kluge’s findings, the differences are notable and should be resolved further. The phylogeny’s margin of error would drastically reduce if there were certain standards. First, because the DNA data in the Gene Bank varies so much even within a single species due to the DNA extraction from different labs, one area of improvement would be if all the DNA extraction and sequencing were performed at the same lab, using the same protocol. This could be done with the current Gene Bank data by not selecting the nucleotide data by length, but rather by the lab that performed the extraction and sequencing. I think this would go a long way to improving the confidence people have in the results. Also it was noted that even with modern computers running 24 hours a day focused on just one program, some of the analysis methods take a long time. To get the best results, one must be willing to spend a great amount of time waiting for the programs to finish. The final conclusion is two fold. First, the Kluge phylogeny is expected to be correct with a potential discrepancy in the Apodora genus. The second is that the DNA phylogeny needs to be preformed again with more time, more DNA sequences from more individuals of more species and with more consistent DNA sequences as well in order to draw further conclusions.
Barker, D.G. and T.M. Barker. The Maintenance and Reproduction of a Little-Known Python, Liasis mackloti savuensis. The Vivarium 5 (6), 1994
Barker, D.G. and T.M. Barker. Pythons of the World, Volume 1, Australia. Advanced Vivarium systems, Inc. Lakeside California, 1994.
Barker, D.G. and T.M. Barker. The New Guinea Carpet Python and the Savu Python – a correct common name. The Vivarium 6 (6), 1995
Barker, D.G. and T.M. Barker. A Tapestry of Carpet Pythons. Reptiles May, 1999
McDowall, S.B. A Catalogue of the Snakes of New Guinea and the Solomons, with Special Reference to Those in the Bernice P. Bishop Museum. Journal of Herpetology 9(1) 1975.
Ross, R.A. and G. Marzec. The Reproductive Husbandry of Pythons and Boas. The Institute of Herpetological Research, 1990.
Schwaner, T.D. and H.C. Dessauer. Immunodiffusion Evidence for the Relationships of Papuan Boids. Journal of Herpetology, 15 (2), 1981.
Work done by A.G. Kluge of the Australian Museum, 1993.
E-mail correspondence with R.T. Hoser.
Back to Biology Index