what a species is, or isn't

When I began spreading pollen on roses in 1972, I didn’t even think about rose species. But gradually I learned about R. spinosissima and other species. And I learned (incorrectly) that species are unchangeable things.

But since that concept did not account for obvious differences in how they grew and in their flowers and their ease of breeding with other roses (etc), my concept of “species” became less strict.

Anyone who examines a group or population of plants that are supposedly of the same species will see considerable variation within the population, just as we see within the species Homo sapiens. After all, the concept of species is a human construct, something that people have created to help them simplify reality by classifying/grouping the phenomena they encounter. Maybe our efforts with roses are not as crude as the generalization that all people with red hair have hot tempers, but they surely are over-generalizations.

Today I came across an article its (URL is below) that deals with concepts of species. I think you’ll learn something valuable by reading it.



Thanks Peter, that is a great article, simple for the layperson to read.

Thanks, Peter. I agree that variability within a species is an important concept to remember when breeding. Significant differences between plants of the same species can occur. On top of that are the possibilities for interbreeding within a particular supplier’s seed plots, as well as the potential for mislabeling or misidentifying at any stage.

My R. carolina and R. virginiana both came from Lawyer Nurseries in MT. I ordered a bundle of 25 or 50 of each, because those were the minimums, and selected the five least thorny plants from each to plant out. The differences between the two species in my field are subtle at best, although the R. carolina seem a little healthier overall. I wonder if Lawyer might have a patch of R. carolina and R. virginiana in their field from which they collect seed every year. Are the two patches close enough together for bees to be flying freely between them? Who identified the species to begin with when they acquired them?

I have started masses of R. carolina and R. virginiana in outdoor seed plots and selected at the end of one season those plants that seemed the healthiest and/or had the fewest thorns. In this way I hope to ensure that the species genetics that I incorporate with modern roses are contributing the best possible genes from their species.

I’ve pondered this question for a while now and this thread seems like an appropriate place to ask for a general consensus:

How many generations removed from a full-species can a hybrid be before we no longer label it a hybrid of that species? For example, if I have a 5th generation plant descended from R. palustris, with no backcrossing or other species crosses in its lineage, can I really refer to it as a “hybrid palustris”?

My personal opinion is 3 generations.


Hi Jonathan, may I ask what crosses from the “original” R. palustris have been made, this might determine the class the third/fourth/fifth generation could be called. I am no expert just asking what comes into my head at the time.


No specific crosses, I meant in general. If you continue to “dilute” a species with genes from modern roses like the hybrid teas.

background reading:



Much depends on the species involved, and whether a particularly obvious trait is strongly associated with the species, and is present in the later progeny. This is particularly true of “Rugosas” that are far removed from Rosa rugosa, but still have the rugose leaves.

The real R. rugosa Thnbg. Is a once-bloomer bearing solitary blooms (rarely pairs). The leaves are small (leaflets scarcely more than one inch long). But the rugose character is expressed in some progeny, despite many generations of “dilution”. In this case one of the most distinctive traits of the original species happens to be transmitted as a unit character.

If the closely allied R. nitida were used instead, the story would be very different because the leaves are not so distinctive.

My R. carolina and R. virginiana both came from Lawyer Nurseries in MT. I ordered a bundle of 25 or 50 of each, because those were the minimums, and selected the five least thorny plants from each to plant out. The differences between the two species in my field are subtle at best, although the R. carolina seem a little healthier overall.

Among several papers on the exact subject of the relationship of American species including virginiana and carolina this is maybe the most relevant though there are a couple other papers as well:

Delimiting Species Boundaries in Rosa Sect. Cinnamomeae (Rosaceae) in Eastern North America

The take-away is what Peter has said, these species exist in a phenotypic and geneotypic continuum with large overlaps in range and genetics. Ploidy plays a role in genetic isolation as well.

Earlenson also had quite a collection of American species early on and addressed the topic too, possibly Karl King can dredge up something on it.

I have started masses of R. carolina and R. virginiana in outdoor seed plots…I hope to ensure that the species genetics that I incorporate with modern roses are contributing the best possible genes from their species.

Hopefully the patches are surrounded by concrete walls to a depth of twelve feet with incendiary fail-safe devices all around.

My notion of species changed in the late 1970s. The goldenrods were in bloom, and as I walked among them I saw some very distinct differences. One type in particular had a thicker stalk, and was stiffly erect. I visited a library to find a book on local plants, hoping to learn about the Solidago species. The author wrote that goldenrods hybridize so freely that the lines between the species have become blurred.

Hmm. This statement made me understand that the “species” is largely a matter of faith. Species cannot be precisely delineated because the (assumed) lines of delineation have become blurred. Why should I accept such an assumption? If species evolve by natural selection of the “fittest”, how does it happen that various types of goldenrod can co-exist in the same field?

Well, others have had similar thoughts long before me. Bateson (1912) observed Veronica spp. that co-existed as peacefully as the goldenrods.

Speciation does not require that a new-and-improve type kill-off all its relatives. And a population can become divided into different “species” while interbreeding. In fact, interbreeding actually hastens the process.

Regarding American roses, it may be useful to consider the American Butterflyweed (Asclepias tuberosa) studied by Woodson (1947, 1953). The 1947 paper has two very instructive maps: I shows the current distribution of three subspecies, and map II shows how the high sea-levels during the Cretaceous, and the ice sheet of the Pleistocene, separated Appalachia from Ozarkia (Ozark plateau plus the Llano uplift in central Texas) and from Orange Island. The modern distribution of the subspecies suggests that subspeciation occurred in these regions.

It is instructive (I think) to compare the distributions of RR. palustris, foliolosis and floridana with those of the A. tuberosa subspecies. A. t. interior has certainly spread further north than R. foliolosa, but the distribution of R. floridana seems more closely parallel, though less southerly, to that of A. t. Rolfsii.

Dr. Erlanson Macfarlane (1966) discussed some of the difficulties in recognizing Rosa spp. She wrote, among other things, “Dr. Boulenger told me that Francois Crepin studied the rose species of the world for thirty years ‘and then went mad.’ Certainly he was never able to publish any revision of the Genus Rosa.”

“Boulenger also stressed the importance of the width of the disc at the base of the stamens and the width of the orifice through which the styles and stigmas emerge. These characters of the flower have been largely ignored by rhodologists, chiefly because they are difficult or impossible to study on dried herbarium specimens.”

I can only wonder whether pollinators distinguish among rose blossoms based on the relative proportions of disc and orifice. The bees I’ve observed among my roses do not seem to distinguish much of anything.

Evolutionary theory has advanced a lot since since we old fossils sat through Biology 101.

The concept of hierarchical species is pretty much obsolete. The DNA is showing us we should be thinking in terms of hybridization networks in which each individual is related to all other individuals to varying degrees.

Hybridization networks are really just an abstract way to quantify and visualize a fitness landscape: Fitness landscape - Wikipedia. The seminal publications on fitness landscapes would be Stuart Kauffman’s works especially “The Origins of Order: Self Organization and Selection in Evolution”:

and another titled “At home in the universe” which got a lot of press at the time (1996):

I’ve tried to come up with a phrase that condenses the concepts of hybridization networks and fitness landscapes in the way that ‘survival of the fittest’ describes evolutionary theory but so far I’ve got nothing. Maybe some of our learned colleagues can help with this?

Just as important, however, is the matter of non-genetic adaptation that, I’m thinking, may lead to a sort of mimicry.

The Evolution of Plant Ecophysiological Traits: Recent Advances and Future Directions (2000)

Lechowicz and Wang (1998) have conducted one of the few studies evaluating phenotypic plasticity in a phylogenetic context (see also Pigliucci et al. 1999). In a study of 16 species of North American spruce growing in ambient and elevated CO2 and low and high water availability, > they found that interspecific variation in many morphological and ecophysiological traits was not associated with the species’ phylogenetic relationships. > However, relative growth rate, which is the outcome of interactions among many ecophysiological traits, showed consistent evolutionary trends across species. Perhaps most interesting, relative growth rate was also less plastic across environments than the many ecophysiological traits underlying variation in growth, but the levels of plasticity in growth rate did not themselves show any pattern of phylogenetic constraint. Evolution of function in extant spruces has apparently involved different patterns of diversification in the mean value of traits affecting growth and in the plastic expression of these traits in differing environmental regimes.

(PDF) Evolutionary significance of epigenetic variation | Christina Richards - Academia.edu
Evolutionary Significance of Epigenetic Variation (2012)
Christina L. Richards, Koen, J.F. Verhoeven, and Oliver Bossdorf
16.5.2 Natural Epigenetic Variation May Be Partly Autonomous
In principle, only epigenetic variation that is not under complete control of DNA sequence variation has the potential to explain phenotypic variation beyond that already explained by DNA sequence variation. It is clear that heritable, natural epigenetic variation is sometimes not autonomous, but under genetic control (e.g. FLC as discussed above). This genetic control can not only be direct, but also indirect through small RNAs, as among-accession differences in DNA methylation can be controlled by differences in accession-specific small interfering RNAs (Zhaietal.2008),which presumably reflect genetic differences in siRNA-generating loci (such as TEs and repetitive sequences). Nevertheless, in the natural population studies described above, correlations between genetic variation and DNA methylation variation were often surprisingly weak. > In > A. thaliana> , for instance, a matrix of pairwise similarities between individuals based on methylation polymorphisms was uncorrelated to a similarity matrix based on genetic polymorphisms at genomewide AFLP markers (Cervera et al. 2002), and patterns of gene-level methylation between > A. thaliana > accessions did not reflect genetic relatedness of the accessions (Vaughn et al. 2007).

Within a population, it does not matter whether a necessary adaptation is epigenetic or “hard coded”. However, when a fair-sized sample of the population is moved to a new habitat, the differences can become apparent as greatly increased diversity.

U.S. Dept of Agric. Bul. No. 159, p. 10 (Sept 28, 2009)
Local Adjustment of Cotton Varieties
O. F. Cook
When a variety of Upland cotton planted in a new district fails to attain the standards of the variety, it is usually very easy to show that inferiority of the new conditions is not the only cause of the failure, or even the chief cause. Comparison of the individual plants with each other will soon make it evident that they are much more unlike among themselves than any reasonable supposition of in equality in the conditions would explain. Many individual plants are likely to be found which have not fallen below the previous standards of the variety.

The best of the plants, rather than the worst or the general average, represent the proper test of the possibilities of the variety under the new conditions. The unequal behavior of the plants will often be found to be a larger factor in the low general average than any definite limitation set by the external conditions. If the best plants are as good as in the home locality of the variety we may have an assurance that the new conditions are not in themselves directly and essentially unfavorable, for in that case none of the plants would be able to attain the fully developed characters of the type. The crop may be damaged as much by changes that arise in the plants as a result of new conditions as by factors that actually limit the growth of the plants, but the nature of the damage is different in the two cases.

These observations may explain some matters of interest to plant breeders. Some traits are handed down as stable units, while others are unstable (flipping in and out even in a single specimen).

Further info on the point of infraspecific diversity.

D. F. Jones (1958) compared the yields of the two series of 317 single and 483 double crosses. Of these, a few of the single crosses achieved the highest yields, but single crosses also had the lowest yields. The double-crosses gave slightly greater yields on average, but the important point is the superior consistency. Wild plants don’t need to fuss about maximum yield.

In the previous note I meant to suggest that diversity within a breeding population (or species) is good for all members. Those that are at their best this year, may be less fortunate next year. While those that are only so-so this year may be the stars next year or the year after.

I never warmed up to Darwin’s notion of intraspecific competition. For him, it was a necessity that drove the species to continual improvement. But there are other ways for a species to change when necessary, while remaining much the same (on average) in the normal course of events.

Erlanson (1934) raised self-seedlings of numerous wild roses, then tabulated the characteristics of the offspring. Based on these traits, some of the offspring could be assigned to different species than the parent.

However, she did not determine which of these offspring would have flourished in the parent’s original habitat. The fact that a plant is heterozygous for some trait cannot be taken as evidence that a plant expressing the same trait as a homozygote would even survive. It sometimes happens that a gene or trait that is harmful or even lethal as a homozygote, may offer a significant advantage as a heterozygote. Of course, a species can get along very nicely even though the majority of the offspring die before maturity.

A plant that is heterozygous for a beneficial pairing of Gg will pass along that same advantage to about half of its offspring (when selfed or crossed with another heterozygote). It doesn’t matter that all the gg seedlings die off sooner later, and the GG are at a disadvantage.

Of course, when breeding with species, we can’t expect to get all the goodies a species has to offer by making a single selection from a single cross.

Erlanson: Sterility in wild roses and in some species hybrids (1931)
“My figures bring out very strikingly the fact that in all the diploid and tetraploid groups on this continent (with the exception of R. palustris) there seems to be a segregation in the factors causing sterility, giving some plants with almost perfect pollen, morphologically, and others with as much as 50 percent aborted grains.”

Erlanson: Experimental data for a revision of the North American wild roses (1934)
“Seedlings of R. blanda from northern Michigan segregated into tender and hardy individuals at Ann Arbor, 300 miles south of their natural habitat.”

I can’t guess what the advantages might be in these cases, but they must exist considering that they survive.

According to Erlanson’s key to North American Rosa species, the prickles of R. virginiana are flattened or absent; those of R. carolina are terete if present. Of course, there are other differences.

Karl, what’s a prickle?

A prickle is something larger than a bristle. Commonly called a thorn.
Check pages 205 and 209 of Erlanson’s 1934 paper for examples of the different forms.

Sounds like “adaptive landscape” Sewall Wright was pushing back in the 1930s.

Proceedings of the Sixth International Congress of Genetics, 1: 356-366 (1932)
The roles of mutation, inbreeding, crossbreeding and selection in evolution

One of the flaws in that school of thought was that Wright and his colleagues spent way too much time imagining the “fitness value” of unit characters. Lots of unit characters … all re-assorting freely. Gene linkage had been discovered in the 19-oughts, but the joy of multiplying and raising 2 to the 1000th power was just too much fun for some folks to abandon.

Lechowicz et al. have studied groups of traits that function together. Two or more groups may involve different genes, and affect the survival of the plants in different ways, but still have roughly the same fitness value.

And beyond these considerations, there is this observation by O. F. Cook (1907):
“There are enough adaptations to occupy many naturalists for many life-times. They can, if they prefer, live and die without hesitating to entertain doubts of the efficiency of environmental causation. And yet the fact will remain that the great majority of the differences between related species and between the individuals of the same species have no environmental utility at all, and are quite unlikely to have had any. This is not to be ascertained by denying or affirming the theoretical utility or uselessness of a few selected characters, but by observing whole orders and classes of organisms to learn the general proportions between differences of characters and differences of environmental relations, and by perceiving that the former vastly outnumber the latter.”

The fundamental differences between Rosa carolina and R. virginiana, for example, cannot be attributed to differences in habitat. As William Bateson (1912) observed, “The theory of Selection fails at exactly the point where it was devised to help: Specific distinction.”

the former vastly outnumber the latter.

This seems to be the case with Malus as well. I can’t find the article but recall reading that all of the various domesticated apples are to be found among the phenotypes of Malus sieversii in the wild, and then some. The Ag station at Geneva NY has a diverse collection of it that they are using in their breeding program. This article I pulled up through google seems to contradict that somewhat in saying that M. sylvestris is also present in the modern apple but my remembrance is that sieversii is the full founder of them all.