My previous post formerly included a brief version of this article, although considering the unusual subject matter and unconventional analysis, I feel this is in need of a more thorough treatment.
Cladistics is a method of classification which uses shared derived characters (synapomorphies) to determine relations. For instance, amongst mammals the possession of wings in the clade of bats (Chiroptera) would be considered a synapomorphy. While cladistics was originally used for determining phylogeny (evolutionary relationships between organisms) and is generally used for that purpose today, it can be used for classification for unlikely subjects like geology and linguistics. The method of cladistics can be used for any hierarchal data set (Young 1995); thus it would appear that the trait information from purported "Many-Finned" encounters would be applicable. In the spirit of Paxton (2009), it should be warned that interpretations of this data need to be limited. The existence of a "Many-Finned" cryptid cannot be proven from this type of data or even suggested; rather, this cladistic data can demonstrate which encounters are indistinguishable from known phenomena/animals and be used to determine how accurate Heuvelmans' classification system was.
So what would the ideal form of analysis for this type of data?
I used PHYLIP as it is free, comes with a variety of programs, and is easy to use - I am not an expert on this subject, so if anybody out there has criticism and suggestions for analyses - PLEASE COMMENT!
Anyways, for my first analyses I used the PARS program as it assumes that ancestral character states are unknown (valuable since I am not concerned with phylogeny), different characters evolve independently, different lineages evolve independently, changes to all other character states are equally probable, changes are a priori improbable over the evolutionary time spans involved (this could admittedly impact the results), retentions of polymorphisms are less probable than those state changes, and rates of evolution in different lineages are sufficiently low so that two changes in a long segment of a tree are less probable than one change in a short segment. As I am unconcerned with all but the best-supported clades, I feel this this type of analysis is sufficient for the data I am striving for.
Since the number of trees recovered will be considerable, I will use the CONSENSE program to determine how frequently certain clades were recovered (out of 1.00 = 100%) in order to determine how rigorously they are supported. I'll be totally unconcerned with clades supported in fewer than 50% of trees and I'll say that 75% is fairly good support, although as these are arbitrary numbers I'll mention some "close-calls".
For the first analysis, only the 26 purported "Many-Finned" encounters will be analyzed, to observe any obvious clusters. I also added an "outgroup" where every character trait was set at "0", as any real organism would be highly speculative and contrary to my non-phylogenetic intents. This is a cladogram which utilizes unknown character states:
Click to enlarge. The only interesting placement is the vague Avalanche encounters outside all other reports.
Here is a tree which results from replacing all traits with unknown character states ("?") with an unknown designation ("0"):
Note that the tree without an unknown designation has far more clades with considerable support, and similar clades would probably also be recovered from qualitative reasoning. From now on, I will be using a 0,1 dichotomy, which is more "honest" in a way because the witnesses it taken at their word. Analyses other than PARS could be used, but they assume a known ancestral character state.
Now, some known animals and phenomena will be included. I will include a wave phenomenon (many humps), a saltwater crocodile (comparison to Vauban), a leatherback turtle (comparison to Hanoi), and several variations of a line of small cetaceans: dlphn1 (line of dorsal fins), dlphn2 (line of dorsal fins, multiple humps, visible breath), dlphn3 (Small eyes, short neck, multiple dorsal fins, pectoral appendages, multiple lateral fins, smooth skin, single hump, multiple humps, long tail, bilobate tail, uniform coloration, contrasting coloration, mottled coloration, visible breath, lateral undulations, vertical undulations). I'll also add Mola mola, which should in theory group nowhere near a "Many-Finned" encounter.
Although it is tempting to do otherwise, only clades with considerable placement should be considered to contain any usable data. It is not surprising that the Leatherback Turtle and Saltwater Crocodile grouped together (marine tetrapods which have the appearance of a dorsal crest and can display a single hump), although it is surprising that the Vauban and Hanoi sightings did not group particularly closely. The proximity of the Osbourne sighting to the marine reptiles is rather telling, and the potential identity as a leatherback turtle should probably be further investigated - I seriously doubt a Nile Crocodile could wind up being that wayward. The inexplicable grouping of the multi-humped Avalanche I encounter with sightings displaying multiple lateral fins is perplexing, although probably due to a sparsity of sampling.
So what happens when additional purported marine cryptids not considered to be "Many-Finned" by Heuvelmans (1968) are added to the mix? Note that this is in fact my first analysis and that prior ones had a number of character states that did not appear at all in "Many-Finned" encounters, although since only synapomorphies are considered this should not be an issue. For this analysis I am including the Valhalla sighting, the Daedalus, the Mackintosh Bell, U 28, Chatham Island (a "Merhorse"/"Caddy"-like sighting), and Mr. Barry's encounter (a "Many Humped" encounter):
Although these cladograms are not meant to be taken at face-value, it can be confidently stated that "Many-Finned" encounters do not form anything resembling a cohesive group and probably represent a multitude of phenomena. It is likely that if more sightings are added, no obvious patterns would emerge except for those highly-detailed outliers. I am uncertain as to how "honest" a quality-control cladogram would be (i.e. removing sightings with a below-average amount of detail) as the vagueness and uncertainty of placement is rather telling of this body of data as a whole.
So how did Heuvelmans arrive at a 9-type classification system? It appears that the majority of his classification was done through "determining characteristics", apparently the ones that show up most frequently in a certain type of encounters. So what happens if we repeat the following cladogram with only traits that Heuvelmans considered to be determining in the "Many-Finned"? I'll be incredibly generous to this analysis and get rid of the un-detailed dolphin descriptions as well as the wave:
Since being outrageously biased towards recovering a "Many-Finned" clade does not fulfill anything resembling its goal, I think it is fair to conclude that Heuvelmans did not objectively classify the purported sightings as he claimed.
It appears most likely that he had an archetype in mind heavily influenced by Aelian/Rondelet's Cetacean Centipede/Great Sea Centipede and the Princess encounter (possibly influenced itself by Rondelet's illustration?) and forced all sightings even vaguely reminiscent into the "type". This sort of classification-as-a-qualitative-art-form approach was probably more socially acceptable in the late 60's, and was certainly a step up from the "mental gymnastics" of Oudemans and the earlier one-sighting-one-type approach by Rafinesque, but we now live in the Age of Cladistics and have a powerful tool for which there is no reason not to use.
Like any other form of analysis, cladistics has the potential to be abused - e.g. like my last example where I attempted to modify criteria to fit a preconceived notion - but if enough work is shown this should not be a problem. Additionally, cladistics should only be used to falsify, e.g. demonstrate that certain sightings cannot be distinguished from known phenomena, and any resultant cryptid-clades should be regarded with the utmost caution. For sightings with considerable detail that cannot be convincingly placed with a known phenomena, a qualitative investigation should probably be initiated, as there are other traits which are best dealt with outside of cladistics (distance, weather, duration, et cetera). I seriously doubt that any amount of data from anecdotal information could be used to demonstrate the existence of a cryptid as the most parsimonious explanation, as only a properly documented carcass can do that, but it could be used as an indicator for a potential new species that warrants further investigation and attempted discovery.
So what happens when additional purported marine cryptids not considered to be "Many-Finned" by Heuvelmans (1968) are added to the mix? Note that this is in fact my first analysis and that prior ones had a number of character states that did not appear at all in "Many-Finned" encounters, although since only synapomorphies are considered this should not be an issue. For this analysis I am including the Valhalla sighting, the Daedalus, the Mackintosh Bell, U 28, Chatham Island (a "Merhorse"/"Caddy"-like sighting), and Mr. Barry's encounter (a "Many Humped" encounter):
Although these cladograms are not meant to be taken at face-value, it can be confidently stated that "Many-Finned" encounters do not form anything resembling a cohesive group and probably represent a multitude of phenomena. It is likely that if more sightings are added, no obvious patterns would emerge except for those highly-detailed outliers. I am uncertain as to how "honest" a quality-control cladogram would be (i.e. removing sightings with a below-average amount of detail) as the vagueness and uncertainty of placement is rather telling of this body of data as a whole.
So how did Heuvelmans arrive at a 9-type classification system? It appears that the majority of his classification was done through "determining characteristics", apparently the ones that show up most frequently in a certain type of encounters. So what happens if we repeat the following cladogram with only traits that Heuvelmans considered to be determining in the "Many-Finned"? I'll be incredibly generous to this analysis and get rid of the un-detailed dolphin descriptions as well as the wave:
It appears most likely that he had an archetype in mind heavily influenced by Aelian/Rondelet's Cetacean Centipede/Great Sea Centipede and the Princess encounter (possibly influenced itself by Rondelet's illustration?) and forced all sightings even vaguely reminiscent into the "type". This sort of classification-as-a-qualitative-art-form approach was probably more socially acceptable in the late 60's, and was certainly a step up from the "mental gymnastics" of Oudemans and the earlier one-sighting-one-type approach by Rafinesque, but we now live in the Age of Cladistics and have a powerful tool for which there is no reason not to use.
Like any other form of analysis, cladistics has the potential to be abused - e.g. like my last example where I attempted to modify criteria to fit a preconceived notion - but if enough work is shown this should not be a problem. Additionally, cladistics should only be used to falsify, e.g. demonstrate that certain sightings cannot be distinguished from known phenomena, and any resultant cryptid-clades should be regarded with the utmost caution. For sightings with considerable detail that cannot be convincingly placed with a known phenomena, a qualitative investigation should probably be initiated, as there are other traits which are best dealt with outside of cladistics (distance, weather, duration, et cetera). I seriously doubt that any amount of data from anecdotal information could be used to demonstrate the existence of a cryptid as the most parsimonious explanation, as only a properly documented carcass can do that, but it could be used as an indicator for a potential new species that warrants further investigation and attempted discovery.
References:
Heuvelmans, B. (1968). In the Wake of the Sea-Serpents. New York: Hill and Wang.
Paxton, CGM. (2009). The plural of 'anecdote' can be 'data': statistical analysis of viewing distances in reports of unidentified large marine animals 1758–2000. Journal of Zoology 279(4), 381-387.
Young, GC. (1995). Application of cladistics to terrane history—parsimony analysis of qualitative geological data. Journal of Southeast Asian Earth Sciences 11(3), 167-176.
5 comments:
You're back! I noted your interruption in blogging, and I had begun to wonder if yet another of the Commetariat had dropped off the face of the Internet.
Quite a blast from the past! I mean, reading your 2007 blog on roughly the same subject, albeit with a different purpose. I give you massive kudos for 1) the effort 2) the skepticism and 3) the subject matter, including Heuvelmans. As you said, it takes a quite a bit of academic guts to even approach this topic. One of my favorite passages in some book on the subject of cryptozoology (it might've been Heuvelmans') described a sea captain who, when told of a sighting of a "sea serpent" remained firmly seated in his cabin. His explanation? If he went and looked, he'd have to chose either to deny his own observations or to be laughed out of serious company. It was much easier to just not look.
Thank you for the links Rich!
The stigma against cryptozoological sightings is a real shame - who knows what has gone unreported? I suppose it is too much to ask for an ideal world that is accepting of such unusual testimony, but at the same time can view it with a critical eye. It is unfortunate that so many reports are vague to the point of critical uselessness, although there are some which are quite compelling - which is why I'm doing this!
I read this recently:
http://forteanzoology.blogspot.com/2009/06/dale-drinnon-amended-cryptozoological.html
Dale Drinnon, who seems the author, boldly stikes of the Great Sea Centipede as simply, VOID
While I am far from assuming the existence of a "Many-Finned" it offends me to see someone so callously close off an area of inquiry.
Yeah, "VOID" is a bit rough. While I'd say the current data suggests mundane phenomena as the most parsimonious explanations, vagueness makes classification difficult, and I'm not sure how probable misidentifications of cetaceans at 100-200 meters are. I recall reading one report where an eyewitness confirmed a line-of-fins to be a pod of dolphins, but I could not find it for this article!
At the same time, I think we can declare Heuvelmans' conception of a hyper-bizarre Archaeocete "VOID" unless some reports which actually have those qualities, ahem, surface.
Post a Comment