Genetic similarity does not prove that organisms are a single species, since the genes we examine are only a tiny subset of the total genetic information. If consistant heritable morphological differences are present, those differences are also genetic, since they must be in order to be heritable. It is just that we are measuring indirectly and as a set of combined and unidentified genes. One only need a single piece of heritable difference to demonstrate lack of free exchange. Shared similarities only indicate a lack of divergence in particular traits since the last interchange. Even if the last interchange was recent, the presence of differences shows that such interchange is not free or constant.
Ecological niche modeling is an interesting recent technique to mention, since it often deals with organisms which are initially considered a single species with relatively little divergence in genetic markers or morphology. What ENM often demonstrates is that, despite low divergence in morphology and genetic markers, certain populations are uniquely adapted to fairly narrow environments. Adjacent groups, despite their similarities, are not as flexible as assumed and are NOT adapted to freely move into each others habitats. So despite a lack of other differences, these groups have inherited environmental adaptations distinct from one another and are isolated in that way. Examples include the
Plethodon fourchensis complex,
P.ouachitae complex,
Aneides flavipunctatus complex, and
Lampropeltis getula complex. Each of these was found to consist of multiple species, each of which is uniquely adapted to habitats measurably different from the others, and each was largely absent from overlapping or intermediate habitats.
To the rhetorical case you mention. You would seem to be describing a situation in which the various taxa are not ecologically isolated, but are isolated by environmental barriers and thus experiencing drift, with the differences yet shallow enough the endemic reproductive barriers do not yet exist [ie, geography prevents interbreeding, not biology]. Prior to mergence, at least some of these populations could be considered valid and separate species. Following mergence, the hybrid swarm itself could be considered a new species, although it could be argued that it is a reconstruction of the ancestral species as well. If your genetic markers don't jibe with your morphological traits, you may need to find new ones which do.
There are a number of similar real world examples, especially among salamanders. There are many cases where mtDNA and nDNA demonstrate different ancestry. Where the mtDNA is not especially unique, it may be an example of hybridization or introgression following hybridization. Where it is further unique from either parental population, it may be a case of hybridization producing a new and now independant species. Taxa for which this is common include
Batrachoseps and
Batrachuperus. Also notable is
Plethodon teyahalee, for which the nomenclatural discussion has focused on the rules regarding hybrid origins [an F1 hybrid cannot be a species type; but subsequent generations can be.
P.teyahalee has an extensive history of hybridization and likely arose by hybrid origin. The type specimens are not F1 hybrids, and thus the replacement name
P.oconaluftee is unecessary].
There are additional cases in which hybridization has been a result of environmental change, either natural or induced by humans. Parthenogenic whiptails (
Aspidoscelis) typically occupy the ecotones between the parental species, and are the product of hybridization. Northern painted turtles have been suggested to be a single merged species, produced by three sibling species re-merging following glaciation. I would have to review that hypothesis, since I know that the most recent data shows a very shallow and rapid mtDNA divergence which might argue against that scenario. European firebellied toads are hybridizing extensively, but primarily only in man-made ecotones between the species, and without either hybridization or introgression much beyond that zone.
Salamandra salamandra terrestris and
S.s.salamandra are morphologically distinct, but seem to be gradually merging following the isolation which established their differences. There is much natural hybridization in
Anaxyrus toads. A number of recognized species are believed to be peripheral isolates of more widely distributed species (
A.houstonensis, A.baxteri), but conversely a number of "species" are now thought to consist of multiple cryptic species. This could in part be related to the problem I mentioned earlier, of paraphyly and polyphyly not being valid reasons for species recognition - The recognition of
A.houstonensis, A.baxteri, Lithobates maslini, or
Dendrobates azureus does NOT compell us to further divide their closest and more diverse relatives (respectively
A.americanus, A.hemiophrys,
L.sylvaticus, and
D.tinctorius. However, the evidence of species level diversity in
A.americanus and
A.terrestris comes from independant sources as well.
Tiger salamanders are still problematic. There are many examples within this clade of morphologically identifiable species which form distinct and sometimes deep genetic subclades. There are also large groups which are genetically diverse but lacking in apparent reproductive isolation. This might be a classic case of paraphyly and polyphyly, with many Mexican species being isolated and distinct but of low variability, a few being hybrid swarms, and a few widely distributed and adaptable forms being highly diverse genetically but homogeneous in breeding and morphology. The evidence will have to be examined case by case, with a wary eye to the impermanence of many of the geographic "barriers" in arid-adapted caudates. Barring introductions of allochthonous specimens, I would suspect many populations isolated by mountain ranges to be valid species, and many of these have already been named and recognized. Such populations account for basically everything west or southwest of
Ambystoma mavortium mavortium and
A.mavortium melanostictum, plus at least one unnamed population currently allocated to the former. A closer eye will have to be applied to populations between the Mississippi and the Rockies, as there may be little or no barrier to gene flow in this region, but isolated distinct pockets can't be ruled out (could Devil's Lake tigers be distinct, while others are simply gray-like blotched/barred tigers?).
It's hard to generalize. Because there are no absolute rules, and nature wouldn't obey such anyway, we have to examine such borderline examples on a case by case basis. From the details you provide, I would conclude that you are starting with multiple valid species, and ending with either a single already-named species [children re-merged with parental populations] or a single new species [sibling diversity combined into a single breeding group].
I think what I failed to clarify enough in my examples is that based under current taxonomy, the homogeneous groups are considered non-homogeneous because of the morphological rather than molecular classification.
That part, I got. It's not especially important. Inherited morphology is molecularly determined by genes subject to selection. "Molecular classification" is determined by genes which are largely immune to selective pressures and thus mutate slowly, like clockwork. The former identifies adaptation, while the latter only identifies gradual accumulation of changes. Anything heritable can identify species status. One fixed difference is all it takes [in principle]. Non-differentiated traits do not disprove a status identified already by differentiated traits. While species are defined by shared traits, they are distinguished by differences. It is therefore easier to prove distinction than conspecificity. It is easy to prove a one- or multiple-feature difference than it is to prove that no features differ. The latter is basically what is needed to prove populations as a single species.
and I totally reject the nonsense about %DNA differences, since that defines some things which routinely and ongoingly reproduce as part of the same population and have free gene exchange different species.
This makes no sense. You cannot have it both ways. Those "%DNA differences" cannot exist if you have "free gene exchange". They can only exist if that gene exchange is limited. That limitation can be post-reproductive, as in the form of lack of hybrid vigor. For example,
Dicamptodon tenebrosus and
D.ensatus interbreed regularly in ONE coastal stream, but their genes are only introgressed into two such streams nearby. While they are able to interbreed freely, gene exchange is not "free" and is confined to the actual zone of contact.
I can't fully accept the traditional definition though, because there are glaring exceptions
Nor should you. This so-called traditional definition is wrong, and as I already pointed out, a fallacious argument. The correct argument is that inability to interbreed proves organisms to be separate species. The converse argument is logically incorrect because the two are not mutually exclusive choices. An hypothesis and an alternate hypothesis must be mutually exclusive. I would argue that this so-called traditional definition is simply a common misconception. The actual definition is based on INability to interbreed, and is correct. The fallacy is formally "false dilemma" and "denying the antecedent".
There are cases where two distinct species hybridise and the hybrids form new species. This is well documented in sunflowers as well as several others, although it is a very rare phenomenon
It's actually very common. I could provide an extensive list of plants, and I don't even know the plants all that well. In herptiles, many species of
Darevskia, Aspidoscelis, Kentropyx, Lepidophyma, and various others are parthenogens of hybrid origin.
Hyla chrysoscelis, Plethodon teyahalee, and a raft of others are sexual species of hybrid origin.
Many popular modern definitions of 'species' immediately count populations as different species the very moment they are geographically isolated. If you were to introduce a few on to an island, they would suddenly be defined as a different species. To me that's just stupid; surely if the only thing blocking gene flow is geography and no differences have *yet* formed, they're not *yet* different species, so geographic isolation is irrelevant.
I agree with your viewpoint, but not your opening statement. While it does happen, I think that isolated populations are usually described as species IF they demonstrate differences from related species. While it's true that such differences are sometimes proven illusory, I don't think it's true that it's common to recognize species solely based on isolation. The late Joseph Collins wrote several times with regard to isolated named subspecies, that these should be recognized as full species because they do not interbreed with others, OR they should not be recognized at all, because they cannot be distinguished.
A very big part of the push towards people being 'splitters' rather than 'lumpers' is the desire to publish and label rather than it being correct or appropriate.
I'm not sure that's really true. Most of what I have read [which is a LOT] concerning new or cryptic species, the splitting is usually well-supported by a suite of features endemic to one population and not another, including reproductive isolation [biological, not geographical]. There are few cases in modern times in which lumping has been a good option. The only instance that comes immediately to mind is the synonymization of
Ranitomeya biolat and
R.lamasi with
R.sirensis. Each was initially described on the basis of a small number of highly distinctive color forms from different localities. Additional samples showed all three to overlap in color and pattern, and to have a common and undifferentiated gene pool. Most cases of lumping involve artificially combining populations which each have unique and independant sets of traits and which do not freely interbreed. It really comes down to a refinement and standardization of what a [sexual] species really is, and realization that all the species concepts of the past had one basic feature in common which is usually readily identified by genetic testing: conspecifics share gene pools which are not shared with heterospecifics.