As you point, separating taxonomic units may be difficult in some cases; and even impossible if you only count with morphological evidence. 

Then, if delimiting taxa is not the main goal of your work but you need it for other purposes instead (as for example, calculate species richness) then I think that the advisable is to be conservative on this respect. That is registering all the variants as the same species but specifing the variants. Thus, the morphological information will be there but if somebody in the future solve the actual taxonomic status of the morphological variants you found, then he/she does not have to amend previous wrong/artificial delimitations.

The question cannot be answered in general terms without knowledge of presence/absence of reproductive isolation: some varieties are very clearly distinguished morphologically, so that superficially look like representatives of different species or even genera, while some ("cryptic") species are almost indistinguishable. Generally:

sympatry and lack of reproductive isolation: variety;

sympatry and reproductive isolation: species;

allopatry/parapatry and morphological overlap exceeding 25%: variety;

allpatry/parapatry and morphological overlap below 25%: subspecies;

allopatry/parapatry and no morphological overlap: species.

So, in fact only in sympatry concrete information as to the reproductive relations is needed, but even here it may (albeit with great caution!) be deduced from presence vs. lack of morphological overlap between, and degree of morphological variability within, the respective forms.

Thanks Roman for your answer.. You state that in allopatric populations where there is no morphological overlap they should be regarded as species but in others you mention an overlap of above or below 25%. Thiis will of course depend on the number and type of characters chosen. Am I correct and if so the more characters chosen the lesser the degree of overlap. This will again lead us back to the question. Will it not?

Amadon's "75% rule for subspecies" refer to the distinction between subspecies and variety - not for species: for this anyway 100% difference is demanded in "phylogenetic" approach and even more in "evolutionary" (Mayr's)- criteria (which demand "weighting" of differences on the "scale" observed in sympatry). So if any character in 75% (or more) of individuals in one population falls out of the range of variability of the other (i.e. the overlap does not reach 25%), these populations should be considered (under Amadon's rule) taxonomically different (i.e. subspecies), otherwise you have to do with individual varieties.

Roman has given you valid set of criteria (not always available for you to judge the decisions made by earlier workers when you use literature!). When you are dealing with sympatric, or even more, syntopic (in the same locality/sample forms, there is certainly NO validity in using subspecies. In some cases there are simply morphs that interbreed freely; in others, detailed examination reveals cryptic sister species. There are plenty of cases where "species" have been split after closer examination. Molecular studies can help here.

When you are dealing with geographically distinct forms (with disjunct distributions, so there are not intermediates) which happens quite often with relatively poor-dispersing invertebrates, or those dependent on a particular and patchily-distributed habitat, you should bear in mind that not all will have named forms using rigorously quantitative methods. In some cases, you will find a multitude of isolated populations described as subspecies; others will have regarded a similar pattern as a set of species; Still others may or may not give you some morphometrics, but decline to give names to the particular populations.

Classic taxonomy works well in deeper nodes where species no longer interbreed.  For cases with speciation in progress, DNA, ecological, and behavioral studies are needed to understand why, from who,  and how much the populations have diverged. We often don't have those studies in hand while needing to give a name to a new form.  So, we use classic taxonomy, but need to keep in mind that the name and placing of that form may change as new studies come along.  I see the use of morphological taxonomy in species that are being formed as a temporary placeholder.  

 "Classic taxonomy works well in deeper nodes where species no longer interbreed" - the "species" which do regularly interbreed are not species - they are either (parapatric) subspecies or (sympatric) varieties!

"For cases with speciation in progress, DNA, ecological, and behavioral studies are needed to understand why, from who,  and how much the populations have diverged" - yes, of course you are right, but these are questions concerning evolution, "phylogeography" &c., not taxonomy! Species are (by definition!) reproductively isolated, i.e. do not regularly interbreed in nature, irrespective of their DNA sequences or "why, from who, and how much the populations have diverged"!

Well, I see that you, Cristina, work with stick-insects - I do not know very much about them, but as far as I know among them parthenogenetic reproduction is frequent, what can "blur" the very meaning of terms like "species", "subspecies" or "variety"...

But, Roman, the problem is that we have no direct evidence as to breeding; we have two(or more) forms that differ morphologically. If they are syntopic, it is Ok to regard then as species. If not, the element of judgment returns.

If PSC (Phylogenetic Species Concept - in fact not "concept", only a criterion, but this here irrelevant) is applied, then 100% diagnosability (between allopatric populations) makes any subjective judgement inapplicable; it is needed if Mayr's version of BSC (Biological Species Concept) is applied, which accepts as taxonomically relevant all kinds of reproductive isolation except geographical. This, in my opinion, is the main (and in fact senseless...) weakness of BSC, therefore in case of allopatry I strongly prefer PSC, and generally what I proposed e.g. in the final chapter of my "Philosophy..."-paper":

"I advocate (and follow in my works – for more detailed discussion see Hołyński 1992) three complementary definitions of species:

(1) evolutionary concept [a modification of Simpson’s (1951, 1961) evolutionary definition]: “species are segments of phyletic lineages (ancestral-descendant sequences of populations), whose gene pools have differentiated beyond limits of reversibility”; this is the most general formulation, closest to the basic tenets and applicable in principle to all groups of organisms, but transgression of the limits of reversibility is very seldom directly demonstrable in actual cases, so usually we must try the second-choice

(2) reproductive criterion [an adaptation of Mayr’s (1940) “biological” definition]: “species are groups of interbreeding populations, reproductively isolated by intrinsic mechanisms from other such groups”; among allopatric, allochronic, parthenogenetic, &c. populations interbreeding obviously does not occur, and the proof or disproof of intrinsic reproductive isolation is either [almost] never possible or irrelevant, so this definition can be actually applied to co-existing bisexual forms only; however, even in overwhelming majority of such instances we have no data on the reproductive isolation as such – this can only be inferred from the observed phenotypical (usually morphological) hiatus, what in fact means recurrence to the only universally serviceable

(3) morphoevolutionary diagnosis [a re-formulation of Hołyński’s (1977, 1992) “operational” definition]: “species are discrete groups of populations, showing – at least in one class (sex, caste, developmental stage) of individuals – consistent unique combinations of morphological characters”.

For less differentiated populations Amadon’s (1949) rule should be applied: if more than 75% of specimens are determinable, we have to do with a subspecies; if less, the form at issue does not warrant taxonomical recognition at all."

I agree with Cristina.

Hello Ahmed,

I think this is a great question that hits on two important questions; first what is a species and secondly what is a subspecies.

First off, to me (and I think may others would agree) the best definition of a species is that of the evolutionary species concept (as Roman explained) where species are "segments of phyletic lineages whose gene pools have differentiated beyond limits of of reversibility". The problem with this is that it is not very applied and thus requires the use of recognition concepts to discern species. The two most popular of these is the biological species recognition concept which focuses on reproductive isolation and the genealogical concordance phylogenetic recognition concept which focus on concordance across multiple lines of evidence (often focus on multiple gene phylogenies). There are issues with both of these concepts.

The biological species recognition concept can only be applied to sexually reproducing organisms (making it hard to use on most of the tree of life, including many multi-cellular organisms) and assumes that hybridization, especially when coupled with introgression, is evidence of reversibility of the lineage. As we gain more and more knowledge about introgression however it is now known to occur without dismantling species boundaries (Fontaine et al. 2015), and likewise, with a rather precise transmission of adaptive traits (Dasmahapatra et al. 2012, Nadeau et al. 2012). Consequently, these results have led to a more accepting view of introgressive hybridization as an evolutionary process, one that instead promotes gene flow across semipermeable species boundaries, and has a precise effect on the genome (Nosil et al. 2009, Michel et al. 2010, Harrison 2012). Thus, hybridization and introgression does not necessity reversibility of species since it does not always equate in fusion.

Recent literature has really started to take a liking to the genealogical concordance phylogenetic recognition concept, especially with the use of NGS data. However, recent authors have questioned its application and the propensity for over splitting of species especially in complex systems and when only genetic data is used. Thus, recent authors have argued for the need of multiple lines of evidence in species delimitation, including morphological, genetic, ecological, behavioral, etc. This to me is the truest way to get at the evolutionary species concept. Below are some papers that do a much better job at explaining thus and showing its application.

https://www.researchgate.net/profile/Bryan_Carstens/publication/249646273_How_to_fail_at_species_delimitation/links/54a152490cf256bf8baf66f7.pdf

http://eherp.org/pdf/88517.pdf

However, the problem with this is that we typically do not have all of this information (although many non-model systems are quickly gaining these resource at an amazing fast rate). So what do we do if we are just a morphologist, ecologist, or geneticist and do not have multiple lines of evidence. To me this is where the use of subspecies, variants, and evolutionary significant units come into hand. The use of these indicators (subspecies and variants often used for morphology and ecology and ESU for genetics) can be very useful to indicating important populations that do not have multiple lines of evidence on there reversibility. Although I do hesitate in supporting the idea of subspecies (see Wilson and Brown 1953) especially given the dangers these can have on conservation when held to a strict sense (see Zink 2004 for an example in avian conservation), I do see its use as Wilson (1995) later recanted "the subspecies category is often a convenient shorthand for alluding to important populations even when their genetic status is ambiguous." In the same sense the use of ESUs can have the same issues when only genetics are used and is especially dangerous given the propensity of over-splitting in new "species delimitation" methods used for NGS data and DNA barcoding (see the two attached papers). Thus, it is in my opinion that the use of subspecies, variant, and ESU are useful and important for research but should be held flexible as more data is gather. Below is a good post about E.O. Wilson's evolving thoughts with some literature cited that you may find interesting.

http://cascabel.typepad.com/cascabel/2015/02/edward-o-wilsons-recantation-on-subspecies.html

I hope this helps and is not just me ranting about species. I just wanted to chime in with my thoughts and look forward to reading the responses to the question.

Best Regards 

Article How to fail at species delimitation

Thank you all for your very critical and enlightening responses and especially to those who have provided  elaborate answers with references. I have a limited sample ( just two specimens) from iIndia which resemble a couple described from Mauritius. They are very close but with what I consider siignificant differences. Since the two localitis are distantly apart and the group itself directly developing I am inclined to consider them as two separate species or subspecies in the process of speciating. I am inclined to consider varieties as synoptic or sympatric populations.

I suggest, pollen morphological characters are most convenient and reliable classical taxonomic tools in deciphering such closely related taxa, before we proceed for DNA analysis which may be essential only in extremely complicated instances. Pollen morphological features vary even at forma level, although in their secondary and / or tertiary characters.

Distinction of taxonomic level can fuel endless discussions.

To be practical, as  a botanist I consider as subspecies populations which all differ from the type by the same set of several constant characters and are encountered outside of the range of the nominal subpecies. Two subspecies cannot be sympatric and within a population only one subspecies is represented.

Varieties are groups clustering indivuals which differ by one or few characters and are encoutered scattered among un-mutated individuals, for example white-flowered or more hairy individuals.