While calculating the evolutionary divergence as computing pairwise distance, (using p-distance methods and Maximum Composite Likelihood model), the values ranged between 0.002-0.138 (p-distance) and 0.001-0.109 (ML) for between individuals of the same species, and for different species. I have got these value as distance matrix table using MEGA 7 software by inserting the nucleotide sequence of COI gene of 661 bp length. Can somebody tell me what are the actual threshold values of the evolutionary divergence calculated by the above said methods to delimit the distinct species? I mean at which values of these distances we would say that these two species/individuals are of distinct species?
I would emphasize even more strongly than the previous respondents that speciation is a process where subpopulations establish their own independent evolutionary trajectories, and any notion of a threshold difference is only a rough approximation at best. As an example, back in the '90s I was doing a postdoc where I was sequencing mtDNA of macaque monkeys. At that time we discovered that rhesus monkeys had major haplotypes differing by about 3%. A very well known evolutionary biologist from the American Museum of Natural History assured me that we had made a mistake because the threshold for species elimination was 1% for mtDNA. Since then a different species of macaque has been found to contain mtDNA sequences that differ by about 9%. Macaques have relatively extreme female philopatry, which explains this unusual observation, but it highlights the folly of assuming an arbitrary threshold of difference.
This artificial "treshold" depends on the marker and the taxonomical group you are working on: you should probably be more specific to have the beginning of an answer.
However, never forget this treshold actually... does not exist. :)
You can empirically measure mean genetic differentiations between species you have defined a priori, but in no wise you can use it to systematically "delimit the species". In particular, since you mentioned a mitochondrial gene (COI), note:
- (1) that true distinct biological species can share identical organelle DNA (look in the literature for "introgression/capture events", "mtDNA replacement", "barcode sharing")
- (2) a unique species can possess deeply divergent mtDNA lineages (i.e. with distances typical of species-level differentiation) with a complete admixture of these lineages once you look at nuclear DNA (https://doi.org/10.1371/journal.pone.0134881, https://doi.org/10.1186/s12862-015-0488-9, https://doi.org/10.1163/15685381-00002895, etc.).
Unfortunately, taxonomy is not (only) about measuring genetic distances. At most, sequences divergence can pinpoint "unconfirmed candidate species" (see Padial et al. (2010): https://doi.org/10.1186/1742-9994-7-16). You then need other types of evidence (nuclear DNA, phenotypical/ecological characters...) to validate whether or not "these two species/individuals are of distinct species".
Best regards.
Dear Riaz,
I would suggest to apply the species criterion (“4 × rule”) to delimit species (Birky et al. 2005), which states that monophyletic lineages represent independent evolutionary units when the mean sequence difference between lineages is more than four times greater than the average variation within the lineage.
Although initially designed for asexual organisms, the “4 × rule” method can also be applied to mitochondrial genes in sexual organisms and has become a promising tool for species delimitation. However I potentially recommend the analysis of other variable nuclear markers to confirm the results.
Good luck,
Temim
@Temim Deli and Tommy Andriollo.
Thank you very much both for guidance. It really help me.
I would emphasize even more strongly than the previous respondents that speciation is a process where subpopulations establish their own independent evolutionary trajectories, and any notion of a threshold difference is only a rough approximation at best. As an example, back in the '90s I was doing a postdoc where I was sequencing mtDNA of macaque monkeys. At that time we discovered that rhesus monkeys had major haplotypes differing by about 3%. A very well known evolutionary biologist from the American Museum of Natural History assured me that we had made a mistake because the threshold for species elimination was 1% for mtDNA. Since then a different species of macaque has been found to contain mtDNA sequences that differ by about 9%. Macaques have relatively extreme female philopatry, which explains this unusual observation, but it highlights the folly of assuming an arbitrary threshold of difference.
One percent or 9% all of these are guidelines that have no objective basis for their value.
I eschew bar-coding personally for a number of statistical reasons. It provides an easy rule with no basis in biology or statistics. Judging speciation has always been difficult and controversial because, of course, the evolutionary process is a continuous one.
Although, strictly speaking, DNA barcoding is not meant to define species, there is something called "DNA barcoding gap" (when interspecific variation is greater than intraspecific variation by an order of magnitude), which is used for separation of species - see the paper attached. Perhaps, you can use it as a rough approximation.
However, species determination is a bit more complicated than sequence analysis and would ideally combine several biomarkers including morphological characters. There used to be a software called "Taxongap", which could handle different types of data (not only sequence data) to create pairwise distance matrices and define OTUs.
For bacteria, as far as I remember, the "gold standard" of species determination was DNA-DNA hybridisation (>70% reassociation) - before the wide adoption of 16S sequencing. For many animal groups, however, using mitochondrial markers may not correctly display the relationship between species, since the substitution rate in mitochondrial genomes in some species is lower than that of the nuclear genome.
As has been said, there is no threshold value for distance. You might consider a threshold value for a measure of the raggedness of a frequency distribution of the distances (aka mismatch distribution). You can calculate these in Arlequin (if not in MEGA). The null hypothesis (not supporting two species in your sample) would be a uninomial distribution, and the alternative hypothesis (supporting two species in your sample) is a binomial distribution. However, even there is not support for two species in these data, it may be worth looking a nuclear data in the future. mtDNA simply takes longer to sort among isolated populations. Consider the likely geological mechanism for isolation of the populations. If you are looking at a late Pleistocene event, you probably will need to look at nuclear DNA.
Another thing to consider is that there are over 20 species concepts that have been proposed, and 3 or 4 that are commonly used by biologists. There have even been serious proposals for abandoning the idea of species altogether. A major rift in these ideas is whether or not species concepts should emphasize the evolutionary process of species formation or biologists' ability to identify them. Your decision needs to be backed by a philosophy of what you think species are and how we can know them when we find them. The latter may go beyond molecular data, which may be limited in its ability to detect very recently isolated populations.
Academically, you may want to just go with your adviser's view at first. But, you will ultimately need to research and think about this yourself.
Best of luck!
Allen
Actually, nDNA lineages are slower to sort than mtDNA lineages, but rapidly evolving nuclear markers (e.g. microsatellites) can be useful for applying other tests of genetic isolation (i.e. the Bayesian method in the program Structure).
Allen Spaulding: Thank you very much for a kind suggestion. I am also working on that.
Sorry I didn't see this earlier!
Much as Allen Spaulding has said, whether two organisms belong to the same species very often depends on which species concept you choose. Different species concepts describe different kinds of entities that need not overlap very much. Peterson & Navarro-Sigüenza (1999) showed that the number of bird species endemic to Mexico is 101 under the Biological Species Concept (a silly name for two different concepts, actually: two sexually reproducing organisms belong to the same species if they can have fertile offspring together, or if they actually do so in the wild...), but 249 under the Phylogenetic Species Concept (under which species are the smallest diagnosable clades), and even the areas of greatest diversity are in different parts of the country under these two concepts! Agapow et al. (2004) extended this comparison to many different organisms and areas.
My conclusion is that different species concepts have nothing in common except the word "species". The general obsession with that word is forced on us by rank-based nomenclature, where you can't give an organism a taxonomic name if you don't refer it to a species...
David,
And yet to do away with species names would be most inconvenient in research. Just because the boundaries among species are rarely discrete and often quite a continuous gradation doesn't mean we do not need the concept. It is an abstraction that has a great deal of practical utility and regardless of the various definitions, most people readily agree on most species boundaries.
I said nothing about doing away with species names. Several different species concepts describe kinds of entities that biologists want to talk about and should therefore be able to name. Actually, the fact that the nomenclature system pretends there's only one such kind (whatever it is) prevents us from naming overlapping ones...
In my experience it isn't at all the case that most people readily agree on most species boundaries. Some fields have stronger traditions than others, or a less dense fossil record; that appears to be all. But that, too, is beside my point. My point is:
Go ahead, name species! Please do! Just make explicit which species concept you're using. That's all I ask.
this article may be interesting for you: https://www.researchgate.net/publication/308955895_Evolutionary_history_and_species_delimitations_a_case_study_of_the_hazel_dormouse_Muscardinus_avellanarius
Article Evolutionary history and species delimitations: a case study...
They depend on the problem at hand. However, there exist several papers where many diffierent recomendations are exposed.
- Badges
- Science topic
- Similar topics
- Biological Science
- Evolutionary Biology