A tangent on mammal subspecies


Hello! I got into an email conversation with my mammalogy professor on mammal subspecies that started from my hesitation on accepting subspecies of Virginia opossums. I thought some of you may be interested in seeing what I had to say since a lot of it stems from my time as an iNaturalist curator.

In general, I believe that the number of subspecies are exaggerated for most mammal species. Many of them were described before advancements in phylogenetics and are instead based on minute variations in skull morphology or pelage. While these can be indicative of a unique population of a mammal, they were mostly described from pretty small sample sizes; didn’t take individual variation into account; and contradict conditions of the environment that would suggest subspeciation had not actually occurred. For example, before being subjected to genetic analysis at one point there were ~90 subspecies of brown bears described, but now it’s understood that most of them are untenable and can be divided into just 5 clades (Miller et al. 2006; Hirata et al. 2013).

I believe there are “good” subspecies, and that these subspecies can be supported (and identified) genetically. But even when a subspecies (or even a species) is described on genetic grounds, sometimes researchers compare the individuals from the farthest extremes of a species’ range and don’t consider that the genetic differences between them could be clinal. At the suspected “boarder” between subspecies A and B, there could be no genetic differences between them. It is on these grounds that the IUCN has recently accepted Alces alces and Alces americanus as one species (Hundertmark, 2016) and that there are no valid subspecies of jaguars (Kitchener et al. 2017). Alternatively, sometimes a subspecies is described from a living population that has been greatly reduced due to extirpation. Consequently the genetic cohesion of the population is a product of inbreeding depression, not natural occurrences of isolation. An example would be Florida panthers (previously Puma concolor coryi), which was more genetically diverse prior to 19th century and samples taken from museum specimens indicate they are untenable from Puma concolor couguar (Culver et al. 2000).

I do not think the subspecies described are “good” or at least need to be legitimized, especially Florida opossums (Didelphis virginiana pigra) and Eastern opossums (Didelphis virginiana virginiana). I do not think they have been subject to any phylogenetic analysis of Virginia opossum subspecies, or I could at least not find one. I am personally skeptical of their distinctiveness because: [1] phylogenetic analysis of other widely-dispersed US mammals (cougars, bobcats, black bears, raccoons) with southeastern subspecies have found that they are untenable (Culver et al. 2000; Kitchener et al. 2017; Puckett et al. 2015; Cullingham et al. 2008); [2] there aren’t any boarders or other means of isolation that would prevent D. v. pigra and D. v. virginiana from interbreeding, potentially supporting a huge “subhybrid zone” that would make recognizing them as distinct meaningless; [3] the adoption of these two subspecies are largely derived from Gardner (1973), who did not cite any phylogenetic or morphologic justification for the subspecies; and [4] the one study I could find that critically compared D. v. pigra and D. v. virginiana - a morphological comparison between their pelvises - found “no significant difference in males or females for any absolute dimension of the pelvis or nonpelvic structures (Wilcox-on-Mann-Whitney test)” (Tague, 2003).

I think too many people accept subspecies (and species) at face-value, and may not understand underlying issues that would make them untenable. Taxonomic units aren’t discrete, absolute truths, and treating them as such sometimes have real world consequences for the animals involved. Recognizing Florida panthers as a separate subspecies from other cougars in the US burdens the population by encouraging inbreeding (by adopting the false implication that they are a unique subspecies) or pushing genetic revival of museum specimens for genetic diversity (which would be unnecessary, invasive, expensive). The best thing for Florida panthers would be to introduce cougars from Texas, Mexico, or California to the Florida Everglades to bring some much needed genetic diversity, but that will only happen if the concept of Puma concolor coryi is abandoned. This isn’t an issue for the successful Virginia opossum, but it’s still something I keep in mind.


Interesting. I think this is sometimes true for plants too and conversely many things that are elevated to species level should themselves be subspecies or varieties. The definition for “species” seems rather fluid and inconsistently applied.


A few months ago I attended a talk by Dan Strickland, who studies Canada Jays (Perisoreus canadensis), and is probably the single living person with the most experience observing them. There are a dozen or so published subspecies (iNaturalist lists 10, Wikipedia 9 current and 2 former). Dan said he believes there are only 3 real subspecies, after observing all of them in the wild and examining museum specimens. E.g. a supposed subspecies on Anticosti Island turned out to be heavier, on average, but otherwise physically and behaviourally identical. E.g. one of the real subspecies along the U.S. pacific coast has feathers that change color due to wear over the course of a year, and specimens of another supposed subspecies had all been collected in the spring. After 20 years in the collection, all the museum specimens of the supposed subspecies appeared identical to those collected in the fall.

Listening to that talk was an eye-opener for me. I’m now much more wary of accepting subspecies without knowing what the evidence is. It seems there are a lot of published papers where a researcher got fixated on one feature or another and used it to separate out and name a new subspecies or even a new species. This is partly because there’s no clear definition or procedure to define whether some sub-population is or is not a subspecies, so there’s a lot of variation in what individual researchers believe is necessary.


Same! I generally don’t accept a subspecies designation unless there is good evidence that it is valid, or if they are accepted by our external taxonomic authorities. However, as a mammal curator on iNat, I really can’t do the latter. Neither of our external mammal authorities (IUCN Red List and ASM Mammal Diversity Database) explicitly accept or reject most subspecies. The IUCN does recognize a few, but for conservation purposes, so it’s not an attention they apply to all taxa. Curating mammal taxa below species is consequently more of a gray area. Some users on the site don’t want us to recognize any subspecies, while others see them as valid taxa.

1 Like

It is true that there are many more subspecies of mammals that are still recognized than probably warrant recognition. Most of the original descriptions, based on pelage color variation and small morphological differences, date back to the 1800s or early 1900s and the majority of these forms have not been looked at genetically. In cases where there are good genetic data, some of these forms have been elevated (or re-elevated) to species level. The reduction in number of recognized subspecies is perhaps more evident in herpetology than in mammalogy, but there is a definite decline.

There are a few mammals for which I think IDing to subspecies level is important, in part due to changing concepts of whether the forms are subspecies or species, or because there is genetic evidence supporting the subspecies taxonomy: American Marten (Martes americana, including the western populations that most marten experts now refer to M. caurina), and Bighorn Sheep (including the Rocky Mountain and Desert subspecies) come to mind.

There is one iNat record reviewer who will ignore earlier IDs of mammals at the subspecies level and instead ID the record just to species, presumably based on the philosophy that there are no good mammal subspecies. I would not go that far.