Evolving Thoughts

Measuring extinctions

In 1972, David Raup published an influential paper on taxonomic diversity during the Phanerozoic. In that paper, he estimated extinction rates based on the number of fossil families and genera for the period and before and after. The idea was to estimate the “kill rate” of major disruptions in earth’s history.

Blogging on Peer-Reviewed Research A new paper by Sarda Sahney and Michael J. Benton attempts to do this for the Permian extinction, arguably the biggest of all time. They attempt to reconstruct the “guilds”, or ecological roles communities, of the Permian, and assess the biological diversity in terms of taxonomic diversity.

It is this that I have problems with.

Sahney and Benton use a Linnaean rank as a surrogate for diversity of species, or rather, two ranks: orders and families. The problem with this is that it is unclear how objective such ranks are. Linnaean ranks were even by Linneaus’ own standards, fairly arbitrary – in fact Linnaeus, who had five ranks only, said they were merely an aid to the student and the collector (and a very good aid they were too).

But apart – arguably – from species, none of these ranks represented anything real. Oh, Linnaeus thought kingdom did, but recent work shows us that the kingdoms are just arbitrary bits of the phylogenetic tree marked out because it suits us to do so. Plantae might be real, but Animalia seems to be artificial, as Fungi ought to be included with it.

The particular rank used by Sahney and Benton – family – isn’t even one of the original Linnaean ranks. It was a term of Michel Adanson, a younger contemporary of Linnaeus, and didn’t get included into Linnaean ranks until 1871. So what does it indicate in the natural world?

Arguably it represents some apprehended degree of difference expressed by taxonomists, but this is highly suspicious when discussing animals. Humans evolved to identify certain kinds of taxonomic resemblance, but the payoff was survival, not taxonomic exactitude. So we are very good at identifying “doggies”, “cats” and “horsies” but less good at identifying, say, deer (the name “deer” is related to the German Tier, which means “animal”).

Taxonomists, however, spend their lives learning about the organisms they study, so perhaps there’s something that is implicit or tacit that they are expressing by the ranks? Some people think that phylum, another rank that was not original to Linnaeus, but came from Cuvier, is “natural”, but at best only within animals. The botanical “equivalent”, division, is not commensurate with the zoological phylum, and neither phyla nor divisions are commensurate with anything in the classification of bacteria, algae and other single-celled organisms.

I have no doubt that there was a mass extinction at the end of the Permian. But the figures for species extinction are bogus, as they rely on these artificial taxonomic ranks. Moreover, there’s no guarantee that even if the ranks were objective, that they would indicate lower rank diversity.

More here at Greg Laden’s Blog, and also at Fish Feet

This is the 700th post here.

Comments

  1. #1 salva
    January 24, 2008

    I’m agree with you, and even if the ranks were originally used by Linnaeus ;)… I’m always disappointed by many papers about extinction and cambrian ‘explosion’ based on rank counts!

    As far as I know the only one who provides an explicit ranking was Hennig who propose using geological periods (then is no more ‘there are no new phyla after Cambric’ because the clades originated after Cambric would be more exclusive than Phylum!), he was an entomologist so they proposed rankings fit more or less insect general taxonomy, but produce some changes in vertebrates (birds, mammals and teleosteans becomes families!). Of course that system would not be better for measuring ‘rank extinction counts’.

    As you say, I think that rank taxonomy have the advantage of put some useful navigation landmarks in the tree-of-life. For me, that is the main reason to maintain ranked taxonomy. Unranked taxonomy have the advantage that things like ‘rank extinction counts’ can not be designed. Maybe the better way is mix both kinds of taxonomy, as is used for example in some DataBases like GenBank!

  2. #2 Thony C.
    January 24, 2008

    the name “deer” is related to the German Tier, which means “animal”

    If this is the case then why is the German word for “deer” Reh?

    Just askin’ Guv! ;)

  3. #3 Matt S
    January 24, 2008

    To your objections I add another: paleo-species identification is bias by preservation. Species differentiation dependent on “hard parts” will show up more than differentiation dependent on less (or not) preserved aspects.

  4. #4 Chris
    January 24, 2008

    I thought “Fungi” was actually paraphyletic, with one branch grouping near the animals and another off on its own? I have bugger all faith in molecular techniques at that depth of time, anyway. And I can never even remember all the single-celled Kingdoms. They seem to spend all their time co-opting each other as organelles anyway.

  5. #5 Christopher Taylor
    January 24, 2008

    Fortunately, in the case of Fungi vs. Plantae vs. Animalia, the cellular ultrastructure as well as the molecular data supports our current view of relationships.

    Animals and fungi are closely related to each other – one of the characters uniting them is a singular, posterior flagellum in those cells that have them (only the basalmost fungi retain flagella) as opposed to paired, anterior flagella in most other eukaryotes. Some of the groups previously regarded as fungi such as oomycetes (water moulds) have paired anterior flagella, and are not a member of the animals + true fungi grouping – instead, they appear to belong to a group called stramenopiles that mostly includes unicellular algae, as indicated by the detailed structure of said flagella. Stramenopiles are probably not closely related to the animals + true fungi clade – they may be related to the Plantae, but we’re getting into areas where the support is more debatable there.

  6. #6 Jim Thomerson
    January 24, 2008

    Paleontologists labor with information so very much degraded and fragmental compared to what we neontologists have. I don’t think one could really do a useful species level assessment of paleoextinction among terrestrial animals. Might be able to do it in marine groups where the data is so much beter.

    I also think species are real, and originate in the universe rather than in my mind. Genera and higher levels of organization are created by fools like me. They are subjective in that they reflect our understanding of the organization of nature; however, they are objective in that they result from our examination of data and the inferences we draw from that examination. There is an argument that any monophyletic group is just as real as a species. (I am not completely convinced).

    When I taught Ichthyology, students very early keyed out and learned to recognize the major families of North American freshwater fishes. The first lab practical involved sight identification to family of specimens they had not seen before. Students could do this successfully even though they knew little or nothing of fish phylogeny, and had had little or no experience with differentiating species.

    I think the paleontologist is in a position much like my students; having a pretty good understanding of what’s there at the family (or order) level and a much poorer understanding at the species level.

  7. #7 John S. Wilkins
    January 24, 2008

    There is an argument that any monophyletic group is just as real as a species. (I am not completely convinced).

    I am. They are real in the sense that there is no convention involved in their demarcation (although there may be in the act of demarcating them). But there is no rank for monophyletic groups, and hence no easy commensurability.

  8. #8 kevin z
    January 24, 2008

    Interesting discussion here. By your argument then, 90% of ecology is misleading since many of it uses number of genera, number of family and number of orders, etc. as surrogates for biodiversity. Many ecologists, such as myself at time, use species, or morphotypes (i.e. Genus sp. or Family sp.1, sp. 2, etc.) when species cannot be assigned in time for publication. I certainly agree with you that comparisons at the family level and above is meaningless.

    So the practical question is, should ecologists subsample their data for DNA? If barcoding takes off like advocates say it is, then should the field biologist collect DNA samples from his research organism(s) to determine what the monophyletic clades are, then use only those clades as the data for the “species” variable. Under the phylogenetic definition that is.

    I’m curious, because in practice the phylogenetic species concepts don’t seem real practical outside of phylogenetic studies. But perhaps biologists need to reorganize their methodology to fit the species concept? I wonder if there has been an ecological study using this approach.

  9. #9 John S. Wilkins
    January 25, 2008

    People talk about phylogenetic species concepts, but so far as I can tell, there is none. Instead there are three conventions: Hennig’s, by which a species is extinguished when it speciates into two or more, the Diagnostic, by which a species is whatever lineage can be diagnosed (or is reciprocally monophyletic) and the “ordinary” one, by which a specimen is taken to be a token of a species and acts as a(terminal) node in a cladogram. The latter relies on other criteria for specieshood than phylogeny, usually the reproductive isolation conception.

    I think genera are also meaningless in this sense, given that what makes a genus a singleton set is that the species is (subjectively or phenetically) regarded as too distinct to be included in some other genus, and what makes a genus heavily populated is (usually) the lack of detailed studies to find the discriminating characters – in other words, nobody’s yet proposed an assay that would isolate them into novel genera.

    One guy who I know has done a lot on ecologically-important measures of diversity is Dan Faith at the Australian Museum – he has several measures (including a phylogenetic diversity metric). He also has an ecological diversity metric. Check him out.

    On barcoding – I’m not quite at the stage where I think it is the snare of the Devil, but almost. At best it is a quick and dirty assay for species diversity, but at worst it is entirely misleading. There’s no principled reason to think that COXIII will always covary with species, rather than, say, be often shared across species or polytypic within them. Maybe it’s useful when that has already been established to some degree of confidence, but I haven’t seen any discussion that shows that it is. So far it’s a Checkout Chick Mentality (if you can scan it, then no more work needs to be done).

  10. #10 Jim Thomerson
    January 25, 2008

    Interesting paper by Douglas, et al. in Copeia 2007 #4. Argument that the Panament Rattlesnake should be recognized as a full species. They discuss four species concepts. (1) Biological species,reproductive isolation. (2) Phylogenetic species, synapomorphies. (3) Geneological concordance, agreement between different independent molecular markers. Said to be a composite of biological and phylogenetic. They prefer the idea of (4) genotypic clustering, said to recognize species by morphological and genetic gaps, and to be theory free in that there is no assumption as to how the gaps arose. I am reminded of Steve Farris’ comment, “A similarity is only a similarity; but a difference is really a difference.” Their rattlesnake is a separate species regardless, as it turns out.

    In my own mind, genera should be monophyletic and diagnosable. Lynn Parenti has made the point that a monospecific genus tells us nothing more about relationship than we know about the species. Nevertheless, I have erected two monospecific genera. Lynn is correct that I cannot differentiate between the characters which make the species and the characters which makes the genus. On the other hand, the two species in question do not fit comfortably into any previous genus. In addition, their sister group is another genus. I think if your sister group is a genus, you are a genus as well, regardless of how many of you there are.

  11. #11 John S. Wilkins
    January 27, 2008

    Jim, I have a full list of species concepts on this blog here. What concepts people actually discuss seems to be a matter of personal taste.

    I agree that if your sister group is a genus then so are you. My only point is that discriminating this way is a matter of judgement and not a forced choice based solely on the data.

  12. #12 djlactin
    January 27, 2008

    I think you have to give Sahney and Benton a break here. Of course, you’re correct that taxonomic rankings are rather arbitrary, but they’re the best estimate we have of taxonomic affiliation and its near-reciprocal, diversity.

    Sure, their approach is holey, but it’s a start.

    What (better) alternative approaches can you propose?

  13. #13 John S. Wilkins
    January 27, 2008

    I have no problem with the use of ecological guilds or other ecological data. I have no problem with the use of species, even given the problems with paleospecies identification. Both of those are likely to be rooted closely in the real world. But the objections to the use of higher taxa were raised against Raup and Sepkoski back in the 70s, and we are still using these measures?

  14. #14 Jim Thomerson
    January 27, 2008

    So, what is the reason paleontologists contiue using families rather than species over some 40 odd years?

    Whether one differentiates species, genera, or families, one is stating an hypothesis of relationship. Hopefully the best hypothesis of relationship which can be put forward given available data and techniques. With any luck, a provisional truth with a high degree of verisimilitude is the result.

    When we started working on Austofundulus and related genera, Austrofundulus had recently (1966)been reviewed in the process of describing a remarkable new species. Most of the other Austrofundulus material was from collections in the 30′s and 40′s. Four species were recognized, one of them with two subspecies.

    We made a large number of collections and were able to study living specimes. We revised the genus (1978): put the new species in its own genus; raised the two subspecies to species rank; and synonmized the other two species with one of them.

    We made a number of additional collections and observations and were joined by a DNA worker. In 2005, we revised Austrofundulus again and recognized seven species.

    In each case the workers involved did the best they could with the material at hand, the available techniques, and their personal biases. Who knows how many species will be recognized in, say, 2050? My point is that, whether species are different from genera and familes, our differentiation of them is constrained by available data, available techniques, and our world view.

  15. #15 djlactin
    January 28, 2008

    I have no problem with the use of ecological guilds or other ecological data.

    I don’t think this is feasible. If you can suggest a credible way of (a) defining ‘guilds’ to at least the precision attributable to taxonomy; and (b) assigning bone shrapnel reliably to these guilds, I’d buy your method. But this process substitutes postulated ecological function for postulated taxonomic relatedness.

    Even with how little we understand the taxonomy of long-extinct organisms, at least there is solid evidence of structure and (less so) of phylogeny. The ill-defined (qualitative) nature of the taxonomic family adds uncertainty to the analysis, but I still think it’s the best we have.

    Further, I think that we could regard ‘family’ as a sort of proxy for ‘guild’. (In the sense that members of a taxonomic family tend to have similar ecological roles: think Canidae, Vespidae, Bovidae…). But why add another layer of uncertainty to the analysis?

    But the objections to the use of higher taxa were raised against Raup and Sepkoski back in the 70s, and we are still using these measures?

    I think that if anybody had come up with anything better, it would be in the literature by now.

    (p.s. the fact that Linnaeus didn’t recognize families, is a non-issue.)

  16. #16 David Marjanovi?
    February 23, 2008

    Plantae might be real,

    Er… sure it is, but its rank isn’t.

    but Animalia seems to be artificial, as Fungi ought to be included with it.

    Why? Animalia + Choanoflagellata + those weirdo parasites are together the sister-group of Fungi.

    Perhaps you were thinking of dividing all eukaryotes into Animalia and Plantae? I can’t see how that would work…

    As you say, I think that rank taxonomy have the advantage of put some useful navigation landmarks in the tree-of-life.

    Names put these landmarks, not the ranks ascribed to those names.

    If this is the case then why is the German word for “deer” Reh?

    It isn’t — it’s the word for “roe”. The word for “deer” is Hirsch — and many people don’t even know that roe are just another deer species.

    BTW, in Dutch, “animal” is dier. The d-t correspondence between (High) German on the one hand and the other Germanic languages on the other is regular (trinken “drink”, Teufel “devil”…).

    I thought “Fungi” was actually paraphyletic, with one branch grouping near the animals and another off on its own?

    The oomycetes — those that are close to the brown algae, yellow algae, diatoms, golden algae, and assorted bizarro parasites — have simply been removed from Fungi. The rest is monophyletic, provided you don’t forget to include the microsporidians.

    I have bugger all faith in molecular techniques at that depth of time, anyway.

    Why?

    By your argument then, 90% of ecology is misleading since many of it uses number of genera, number of family and number of orders, etc. as surrogates for biodiversity.

    In fact, yes, though I doubt it’s 90 %.

    Enter the Phylogenetic Diversity Index:

    Faith, D. P. 1992. Conservation evaluation and phylogenetic diversity. Biological Conservation 61: 1 — 10.
    Faith, D. P. 1994a. Genetic diversity and taxonomic priorities for conservation. Biological conservation 68: 69 — 74.
    Faith, D. P. 1994b. Phylogenetic pattern and the quantification of organismal biodiversity. Philosophical Transactions of the Royal Society of London, Series B: Biological Sciences 345: 45 — 58.
    Nehring, K. & Puppe, C. 2004. Modelling phylogenetic diversity. Resource and Energy Economics 26, 205 — 235.

    For that you only need a tree, not ranks.

    I certainly agree with you that comparisons at the family level and above is meaningless.

    Even at the genus level it’s meaningless, because one worker’s subtribe is another’s subgenus.

    If barcoding takes off like advocates say it is, then should the field biologist collect DNA samples from his research organism(s) to determine what the monophyletic clades are, then use only those clades as the data for the “species” variable. Under the phylogenetic definition that is.

    DNA barcoding uses neighbor-joining, which is a phenetic method, not a phylogenetic one: it counts shared similarities rather than just shared derived similarites. DNA barcoding basically imposes a phenetic species concept.

    Also, “monophyletic clades” is double. “Clade” and “monophylum” are synonyms; the definition of “clade” is “an ancestor and all its descendants”.

    There’s no principled reason to think that COXIII will always covary with species, rather than, say, be often shared across species or polytypic within them.

    This depends massively on the species concept.

    After all, depending on the species concept there are between 101 and 249 endemic bird species in Mexico! (Source: Talk by Yann Bertrand at the 2nd meeting of the International Society for Phylogenetic Nomenclature.)

    They prefer the idea of (4) genotypic clustering, said to recognize species by morphological and genetic gaps, and to be theory free in that there is no assumption as to how the gaps arose.

    Proudly phenetic, then.

    In addition, their sister group is another genus. I think if your sister group is a genus, you are a genus as well, regardless of how many of you there are.

    You could equally well have referred the new species to the genus that includes their sister-group. It was a matter of personal taste.

    Further, I think that we could regard ‘family’ as a sort of proxy for ‘guild’. (In the sense that members of a taxonomic family tend to have similar ecological roles: think Canidae, Vespidae, Bovidae…).

    This is completely untenable.

    I think that if anybody had come up with anything better, it would be in the literature by now.

    As shown above, it is. One of them is even published in a journal that paleontologists read.

  17. #17 David Marjanovi?
    February 23, 2008

    Lynn is correct that I cannot differentiate between the characters which make the species and the characters which makes the genus.

    “The characters don’t make the genus, but the genus gives the characters”…

    (Linnaeus as cited by Darwin. The original is AFAIK a bit different, but not much.)

  18. #18 windy
    February 23, 2008

    So we are very good at identifying “doggies”, “cats” and “horsies” but less good at identifying, say, deer (the name “deer” is related to the German Tier, which means “animal”).

    You mean, English speakers are less good at identifying deer :) In other languages there may not be a lack of deer-words as Thony and David pointed out (even to the point of confusion with the different words)

    Actually, in medieval Finnish the lion was called a deer…

  19. #19 John S. Wilkins
    February 23, 2008

    English speakers have a multitude of words for a limited number of ideas… it’s the joint Norman/Saxon heritage.

    Similar “mass nouns” for animals include “cattle” (from “chattel”, personal property). I think it’s probably due to the legal status of animals in English common law.

    I thought Finnish was unrelated to Germanic. Why was the lion called a deer? I presume it’s not derived from “Tier”. And were there many lions in Finland?

  20. #20 windy
    February 24, 2008

    Similar “mass nouns” for animals include “cattle” (from “chattel”, personal property). I think it’s probably due to the legal status of animals in English common law.

    I wonder if it’s more widespread than that (adopting a generic term for this specific group). At least Finnish and Swedish have terms for ‘livestock’ that nowadays imply bovines.

    Why was the lion called a deer? I presume it’s not derived from “Tier”. And were there many lions in Finland?

    It’s because of the lack of lions that no one thought it was strange to call them ‘jalopeura’ (noble or great deer = probably moose!) There are two theories: Either the guy who translated the Bible simply made it up, or that it was mistranslated because the constellation Leo was known by the locals as the constellation of the Moose.