Wednesday, November 11, 2015

Untangling the Viny Viornas: A Case of How We've Underestimated Biodiversity in Architecturally Complex Genera (Part 2 of 2)

In the first part of this series, I laid the groundwork for how architecturally complex plant groups can be difficult to study. Perhaps no group better exemplifies this than the viny viornas or leatherflowers of the genus ClematisLeatherflowers are herbaceous vines that grow up to about 9 ft in length. They generally trail over surrounding vegetation or climb shrubs. Their architectural complexity stems from various floral and vegetative characteristics. Clematis is represented by more than 400 species globally. Since the 1940s between 8 and 12 species have been recognized in the eastern U.S. by taxonomists. Ralph Erickson in his treatment of the viornas recognized 12 species, whereas W. Michael Dennis recognized 8 species in his PhD dissertation which focused on the biosystematics of the group.

I first became interested in Clematis in 2003 during the first year of my PhD program at the University of Tennessee. Even though I lived in Knoxville, I regularly traveled back to central Tennessee to collect specimens. One late spring day I traveled from Knoxville to collect plants in the Nashville area along the Cumberland River. During this collection trip I happened to collect Clematis viorna from moist forested slopes in Jackson Co., TN in the upper Cumberland River Valley. The flowers were bright, cherry red on the outside with a cream interior and most parts of the plant were relatively smooth. The sepals were weakly corrugated along the back. The flower stalks (peduncles) were longer and the bract leaves were larger. The leaves of these plants were biternately compound. Later that same trip, and about 75 miles to the west, I collected another Clematis that also keyed to C. viorna, but was clearly very different when examined side by side. These plants from dry rocky wooded slopes adjacent to limestone cliffs in Cheatham Co., Tennessee had pale-lavender flowers fading to cream sepal tips, sepals grooved on the back, a different leaf dissection, and were soft-hairy on leaves, stems, and flowers. I thought to myself, how can these two collections represent the same species? Examine Fig. 1 and ask yourself the same question!

Fig. 1. The red star indicates the location of specimens of C. viorna collected in Jackson Co., TN (plants in upper right with bright reddish flowers. The blue star indicates the location of specimens of C. viorna collected in Cheatham Co., TN (plants in upper left). Both were collected along the Cumberland River and both key to C. viorna in most keys. The bold green line represents an important geographic break within eastern North America. Many primarily Appalachian species do not occur west of this line and many "western" species rarely occur east of this line. 

So after careful observation in the field and much thought on the subject, I did what any taxonomic botanist would do. I went back to the University of Tennessee Herbarium to examine specimens and to dig into the taxonomic literature to see what previous Clematis experts (Ralph Erickson, Mike Dennis, and Jim Pringle) had to say about variation in Clematis viorna

Features that seemed so clearly distinct in the field such as flower color, grooving of the sepals, leaf dissection, pubescence, and position of the leafy bracts on the flower stalks, all seemed to "break down" upon initial inspection of  all of the C. viorna specimens at UT (see Figs 2-3). Flower color and sepal grooves were impossible to tell because of discoloration of the flower and distortion (through flattening and shrinking) of the sepals upon drying. Leaf dissection was challenging because some collectors collected only the distal-most stem section whereas others collected from mid-stem sections, or in other cases leaves were so poorly pressed that they were sometimes folded and their original dissection pattern was impossible to discern. Thus, leaf pubescence and bract position were the only really "useful" features that could be used when examining most herbarium specimens.  

Fig. 2. This specimen illustrates some of the difficulties of working with Clematis in the herbarium. This specimen is poorly pressed with overlapping leaves and stems and folded leaflets. Compare this specimen with Figure 3 below.
Fig. 3. This is the same species as shown in Fig. 2 but note that the critical features needed for identification are shown, including the dissection pattern of the leaves and the general features of the flowers, their peduncles, and bracts. Collectors of Clematis are encouraged to press specimens with such characters in mind.

Some general patterns seemed to hold when I examined leaf pubescence in Clematis viorna. Most specimens could be separated into a relatively smooth "Appalachian" type and a pubescent "Interior" type. The "Appalachian" type seemed restricted to the Appalachian Mountains, Piedmont, and Atlantic Coastal Plain from northern Georgia and east Tennessee north into southeastern Ohio and Virginia. The "Interior" type seemed to be the main type west of the Appalachian Plateaus (Cumberland and Allegheny) west to the Ozarks and ranging north-south from southern Indiana to central Alabama.   

What was troubling to me was how could these plants be so distinct in the field and then seem to break down in the herbarium? Using hairiness as one criterion to sort specimens, I was intrigued by populations within the Appalachian zone that were quite hairy like "Interior" populations. These isolated "Interior" populations stranded in Appalachia popped up in interesting places like in east Tennessee and southwestern Virginia in the Ridge and Valley Province, near Shenandoah National Park in Virginia's Blue Ridge Mountains, and in the Ocoee River Gorge of  Tennessee's southern Blue Ridge. I only had one or two specimens of each of these outlier "Interior-like" plants to examine so I chalked them up as aberrant individuals and was left scratching my noggin.

Not wanting to get entangled in a leatherflower thicket, I decided to put the problem aside for another day given that I had an actual PhD dissertation to finish -- oh yeah that. Years went by and I had the occasion to study additional specimens from time to time, once during a Christmas break trip to the Missouri Botanical Garden in St. Louis, another when I pulled an all-nighter at the University of Alabama Herbarium on my way to Panama City with the family, and many other trips to eastern U.S. herbaria, including small ones like Stephen F. Austin State University in Texas and the famed Gray Herbarium at Harvard.

During my time in grad school, I added my own contributions to other aspects of Clematis taxonomy by describing a new species from the Carrizo Sands of east Texas, which I named Clematis carrizoensis in 2006. My friend and colleague, Chris Fleming and I also discovered Clematis morefieldii, a federally endangered species, new to Tennessee. 

Fast-forward a few years to 2009 and I again stumbled into the Clematis problem. My good friend and colleague Theo Witsell, botanist for the Arkansas Natural Heritage Program in Little Rock (who I affectionately refer to as my "red-headed brother from another mother from across the Big River"), sent me some specimens to examine (which I still have on loan!). In that batch were specimens of an unusual Clematis from the Arkansas River Valley and adjacent Ouachita Mountains. These specimens resembled those of the "Appalachian" type described earlier in having biternately dissected leaves, long peduncles, and in being relatively smooth. However, they possessed a suite of additional characters which readily distinguished them, including narrowly cylindric flowers, flowers that were light pinkish-purple fading to cream tips, and peduncle bracts that were generally lobed or divided into three leaflets (ternate). My botanical buddy and botanist extraordinaire, Aaron Floden, and I decided to take a trip out to visit the Arkansas populations with Theo and our feelings that it was a distinct undescribed species were strengthened. In the herbarium these Arkansas plants had been identified one of two ways. Some botanists identified them as Clematis viorna -- the name I considered to be the "Appalachian" type with cherry-red flowers. Others called these plants Clematis reticulata, even though they scarcely had reticulate leaves.

For the first time it occurred to me that part of the problem with studying Clematis is that most people had previously relied on herbarium specimens. This makes sense given that a researcher can borrow thousands of specimens from dozens of herbaria and within a few weeks can study a large percentage of specimens ever collected for a particular species. These specimens span nearly two centuries and represent the individual collections of dozens of collectors from potentially hundreds of counties and dozens of states. It would be impractical, likely impossible, for a single researcher to try and see most of these populations live in the field. Just think of how costly such a trip would be with all the expenses of fuel, food, hotel or camping fees, automobile repairs, etc. The ideal taxonomic study should involve extensive herbarium study and ample amount of fieldwork. Usually study of herbarium specimens reveals variation and a researcher can then navigate to those populations to study them live, make photographs, collect tissue samples, and collect material for common garden studies.

Through my experience with examining thousands of herbarium specimens and observing populations from across the Southeast I realized there had to be more to the story given all of the variation in morphology. I decided to dig deeper for clues and again I turned my attention back to herbarium specimens to see what could be leading folks (including me!) astray. Specifically, I wanted to search for specimens that troubled previous botanists. While examining specimens from the University of Arkansas Herbarium with Theo we found what I had been looking for, only it opened the door to a whole new way of thinking. What if taxonomists have been greatly underestimating the number of Southeastern species of Clematis? 

The specimen that led me to ask this question is shown in Figs. 4-5. When this specimen was first collected in 1955 its collector did not identify it to species. Eleven years later, E.B. Smith annotated the specimen as Clematis versicolor. Ten years after that, W. Michael Dennis determined the specimen to be C. reticulata. In 1988, Smith returned with a second annotation and considered the specimen to be C. reticulata introgressed with C. viornaWhy did this specimen receive so many conflicting annotations? What led these experts to label this specimen with three different species names?

Fig. 4. Specimen of Clematis sp. nov. (Ouachita Leatherflower) from the University of Arkansas Herbarium, illustrating the hallmark of undescribed species represented in herbaria.
Fig. 5. Note the lack of lack of a consensus among
experts as reflected by the conflicting annotations
for this specimen.

I said to Theo, "what if there are several undescribed and narrowly endemic species of Clematis from different regions across the Southeast that have clouded our thinking?" Perhaps all along I and other botanists have had trouble with Clematis because we were discounting the potential that much of what we have been looking at in herbaria could be undescribed.  Have we been trying to  "pigeonhole"  undescribed species into known species?

Specimens such as this one tell a story. They can be indicators of undescribed species. They bear conflicting annotations which point to a lack of consensus among original collectors and regional or taxonomic experts. I have seen such specimens in other taxonomic groups and they have frequently been annotated as possible hybrids, just as this specimen was labeled as C. reticulata "contaminated with" C. viorna

According to my own research, this specimen represents an undescribed species. So why did it take until now to realize its distinctiveness? Furthermore, why were past botanists so quick to invoke the hybrid scenario first? Why not simply put a note stating that the specimen does not fit well within either C. reticulata or C. viorna? Or, why not annotate it in any number of other ways, such as calling it Clematis aff. reticulata or Clematis cf. reticulata. It seems that in some groups some botanists are all too quick to jump to the conclusion that an unusual plant is a hybrid, but I rarely see folks going to the other extreme and questioning from the start as to whether a specimen could represent a new species.

I'm not advocating that we should immediately assume that a plant that doesn't seem to fit a known species be considered an undescribed species at the onset. After all, it is a fairly rare event to encounter a new species. Just in the same manner as doctors are trained to first consider that a patient's ailments are due to some common cause, they must also be willing to consider that although against the odds a person could have some very rare condition. I have seen numerous cases where perfectly distinct species were initially regarded as mere aberrations, possible hybrids, or lumped away and considered to represent variation within a morphologically plastic species. This may have been done out of convenience for it often takes a lot of work to determine whether something is indeed a new species or not--as it should. 

It isn't clear why previous taxonomists didn't notice this new species but I suspect that there are probably three main reasons: 

  1. in the late 20th century I think that many botanists were reluctant to believe that some areas of the South could still harbor undescribed species, especially narrowly endemic taxa; 
  2. the fact that our collective knowledge of Southeastern phytogeographic patterns was still in a relatively elementary state in the 1970s-80s due to inadequate collecting in some areas and lack of published studies of biogeographic patterns of Southeastern plants, and 
  3. the fact that Clematis is an architecturally complicated genus that presents numerous difficulties to herbarium study.

First, I would contend that in some cases botanists think that everything has been discovered already or they are reluctant to believe that an unusual population could represent an undescribed species. For a great discussion on this topic see Barbara Ertter's publication Floristic surprises in North America north of Mexico. 

In the 1970s-80s, the Ouachita Mountains had not yet become known as a center of vascular plant endemism. However, since the 1970s numerous narrowly restricted species have been discovered in this ecoregion, including Amorpha ouachitensis (1975), Cardamine angustata var. ouachitana (1982), Carex ouachitana (1987), Polymnia cossatotensis (1989), Hydrophyllum brownei (1991), Sabatia arkansana (2005); several additional species remain undescribed (Theo Witsell, pers. comm.). The fact that this region is a burgeoning hotspot suggests that we should actively seek out potentially undescribed species from the Ouachitas that may be "hiding out" in herbaria. A 2010 publication by D.P. Webber and numerous coauthors titled "Herbaria are a major frontier for species discovery.", published in the Proceedings of the National Academy of Sciences, suggests that many undescribed species are hiding out in herbaria awaiting our discovery.  

Second, prior to the 2000s we did not have a very good handle on phytogeographic patterns of the Southeastern flora. Thanks to major botanical collectors such as Robert Kral, R. Dale Thomas, Delzie Demaree, Roland Harper, and E.J. Palmer, hundreds of thousands of plant specimens have been collected from all over the Southeast. Prior to the 1960s taxonomists had only a few specimens from each state to examine and there were significant gaps with many areas under-collected. This made it difficult to identify geographic distribution patterns, especially those involving narrowly endemic species. Fortunately as species distribution maps were published online (e.g. BONAP) our understanding of plant distribution patterns has advanced greatly. It is now much more widely accepted that we CAN and DO have species that are found in only one or two counties, but I think most taxonomists in the Southeast prior to the 1970-80s considered this unlikely, thus many restricted endemics were considered to be aberrant individuals or hybrids rather than distinct species. Of course there are exceptions. Botanists like J.K. Small recognized many narrowly restricted species. His work with the numerous endemics of Florida probably influenced his tendency to do so. Reed Rollins of Harvard University also is well-known for his work on the narrowly endemic species of Leavenworthia and Lesquerella, which harbor numerous narrowly restricted species including a couple that are single-county endemics. 

Detailed distribution maps like the one in Fig. 6 for Clematis reticulata available from BONAP were not available until recently. Such maps were compiled from specimen data available in dozens of herbaria. When Ralph Erickson was constructing his distribution map of C. reticulata in the 1940s many counties highlighted in the BONAP map below had no collections of C. reticulataWithout the intense collection of specimens during the latter half of the 20th century the many gaps in the range of the species that Erickson encountered would not have been filled. For example, look at the large gap in the range of C. reticulata that separates populations on the east and west sides of the Mississippi River. Phytogeographers now have enough data from many dozens of species (e.g. longleaf pine) that show that this is a very common gap. However, in the early 20th century the paucity of plant collections from many parts of the South made such patterns difficult to interpret whether a species was truly absent from a region or just under-collected. Such inadequate maps made it difficult to determine phytogeographic patterns. In another case, note the two central Texas counties that contain C. reticulata but that are separated from the other east Texas populations. Is this a real gap or would additional fieldwork likely fill this void?

Fig. 6. Distribution of Clematis reticulata according to BONAP (2015).

The lack of accurate distribution maps meant that we did not have a good concept of phytogeography until relatively recently. Robert Kral's two volume publication A Report on Some Rare, Threatened, or Endangered Forest-Related Vascular Plants of the South greatly advanced our knowledge of plant distribution patterns, especially of rare species. By the turn of the 21st century several published atlases of plant distribution patterns had been published. Various publications have focused on distribution patterns of Southeastern plants (phytogeographic patterns), including Bruce Sorrie and Alan Weakley's paper "Coastal Plain vascular plant endemics: phytogeographic patterns" and Doug Zollner et al.'s (2005) paper entitled "Endemic vascular plants of the Interior Highlands, USA."  

I believe that the third major reason for the underestimation of diversity in Clematis is due to their architectural complexity. They aren't like dandelions or other small plants where you can put the whole plant on a single herbarium sheet. Most are long vines, but specimens usually consist of a 10-16 inch section of the vine, and often the distal-most portion of the stem. As a result many potentially important characters are often not collected or documented such as mid-stem leaf dissection. Other features are quickly obliterated by pressing and drying, such as flower color and shape. Below, I discuss some of the characteristics of Clematis and the challenges they present to taxonomists.

The nature of the inflorescence in Clematis has been recognized as taxonomically important but still much confusion exists (Fig. 7). For example, Clematis morefieldii has axillary flowers that usually hang down below the leaves on very short pedicels and that are subtended by small hairy bracts. In contrast, C. viorna  has ascending peduncles that are usually quite long and subtended by two foliaceous and generally smooth bracts. Some botanists have been reluctant to accept C. morefieldii as a distinct species, presumably in part because of the numerous "intermediate" forms that exist between it and C. viorna. For example, populations currently considered to be C. viorna from west-central Alabama and eastern Mississippi and other populations in northern Arkansas are very much like C. morefieldii in their short peduncles. Peduncle length and bract size do vary, particularly with respect to developmental stage. Some species, like C. viorna, tend to have shorter peduncles with smaller bracts earlier in the season at the tip of the stem. Those developing later and farther down on the stem tend to have longer peduncles and larger bracts. With continued development, additional flowers and bracts may develop. Even with this variation on a single plant, the flowers and inflorescences of each entity seems to be constrained  within a certain range.

Fig. 7. Plate showing inflorescence structures of various Clematis. The image at the far left is of Clematis morefieldii. The image at far right is of typical C. viorna. The four center images are various populations currently identified as C. viorna. Note differences in peduncle length, peduncle orientation (ascending vs. descending), bract size, and bract position.

The flowers of Clematis resemble small bells (Fig. 8). Unfortunately, floral characters do not preserve well on herbarium specimens. As a result, important features such as flower shape have largely been ignored by taxonomists. Some species (e.g. C. glaucophylla) have narrowly urn-shaped flowers whereas others are broadly egg-shaped in outline (e.g. C. morefieldii). In some species, the tips of the sepals recurve prominently (e.g. C. viorna) but in other species the tips remain straight and do not recurve (e.g. C. glaucophylla). Another feature that seems to be useful for discriminating taxa is whether or not the backs of the sepals are ridged or smooth. Clematis viorna and C. carrizoensis are examples of species with smooth unfurrowed sepal backs whereas C. reticulata and C. flaccida are prominently grooved. Finally, some species, like C. crispa and C. pitcheri, have sepals with thin, wavy (crispate) margins whereas others lack such a dilated margin.

Fig. 8. This plate shows how fresh flowers can yield potentially taxonomically significant characters that are obliterated during pressing and drying. The column to the far left represents Clematis morefieldii which prior to 1987 was considered part of C. viorna. The three remaining columns are all from populations currently identified as C. viorna. Note the major differences in flower shape, degree of sepal reflexion, sepal tip length, flower color, degree of sepal corrugation, and sepal width.

The general color of Clematis flowers is very important taxonomically (Fig. 9). Some species such as C. texensis, C. carrizoensis, C. versicolor, and C. glaucophylla can only be reliably distinguished by flower color.  Flower color varies with time of flowering and ecological conditions. Plants in full sun typically have lighter colored flowers than those in shade. Distribution of color is also important. Some species have sepal backs that are monochromatic or have one color from base to apex. This is the case in C. texensis, which is scarlet throughout. Other species like Clematis versicolor are bicolored, with darker coloration near the base and becoming lighter toward the tip.  The coloration of the inside of the flower can also be important taxonomically. Some species, such as C. vinacea have sepals with the same color on both the outside and inside surfaces. Others, like C. viorna are dark outside and cream-colored inside.  Most species have a fairly wide range of color variation even within a population or on a single plant, as illustrated in different flower colors on a single vine in a population of C. subreticulata from Mt. Cheaha, Alabama. In spite of the variability of flower color within a species, each species generally has its own particular color range. 
Fig. 9. This figure shows the difference between two plants that were formerly both treated as a single species, Clematis viorna. The two images on the left represent typical C. viorna. Notice the distinct cherry-red exterior and creamy-yellow interior. The two photos on the right represent the recently described C. vinacea. Note this species has flowers with the same color on the exterior as the interior. Also note the difference in sepal pubescence.

Typically, herbarium specimens of Clematis only include the terminal section of the vine, usually the last 8-15 inches or so. Unfortunately, such specimens rarely show mid- and lower stem leaves, which often are larger and more dissected. Most of the viornas have leaves that are compound. The leaflets are not always attached in a single plane to the main rachis, rather they often attach at an angle and are held in a different plane than the axis. Pressing specimens obscures this aspect of leaf morphology. The degree and type of leaf dissection is also important taxonomically (Fig. 10), though such characters have been used rarely to separate species. Some species typically have mid- and upper stem leaves that are pinnately compound and the leaflets often lack lobes (e.g. C. morefieldii) wheras other species are often biternately compound throughout (C. viorna). Careful observation of the variation in leaf dissection should be made on a single plant and across the population.

Fig. 10. Both of these leaves are from plants that would key to Clematis viorna in nearly all plant identification manuals. The specimen on the right represents true C. viorna from the Appalachian region whereas the photo on the left is from an undescribed species in the Arkansas Ozarks.

Past taxonomists have used degree of leaf vein reticulation to distinguish some species of Clematis. This refers to whether or not the veins are raised above the surface of the leaf blade forming a net-like reticulum as opposed to the veins being flush with the leaf surface or embedded in the leaf tissue. Herbarium specimens rarely capture the full range of venation for a population. Plants growing in shade tend to have less reticulate veins and the leaves are often thin or flaccid whereas plants in full sun can have strongly reticulate-veined and thicker leaves. For example, vegetative or fruiting specimens of Clematis versicolor in deep shade are nearly impossible to separate from specimens of C. glaucophylla, whereas sun plants are easy to distinguish based on leaf thickness and venation

Degree of leaf vein reticulation has been the primary character for separating Clematis viorna from C. reticulata (Fig. 11). The former is considered to have veins that are not raised above the surface of the leaf or with only the primary and secondary veins slightly raised whereas C. reticulata usually has primary, secondary, tertiary, and quaternary veins raised above the surface of the blade, forming a net-like reticulum. Like the C. versicolor-C. glaucophylla example above, shade vs. sun leaves complicates interpretation of this character. It is possible that degree of reticulation is due to clinal variation. Populations of C. reticulata in Florida, for example, have very thick and strongly reticulated leaves but as you go northward through Alabama populations of what have been called C. reticulata become thin-leaved and less strongly reticulate. 

Fig. 11. Graphic showing the range of leaf vein reticulation from non-reticulate leaves at the far left to strongly reticulate leaves at the far right.

Restoring Order to the Viornas

My recent research suggests there may be as many as 9-12 previously undescribed species of leather flowers in the Southeast (Figs. 12-13). All of these are being "carved out" of either Clematis viorna or C. reticulata, both previously considered widespread and variable species. Two species have already been separated in recent years, including Clematis morefieldii and C. vinacea, both split out of C. viorna from the Southern Appalachians.

Fig. 12. Overview of the Clematis viorna complex. Typical C. viorna is represented by the red circles with yellow border in the Appalachian region. The stars represent species already described (C. morefieldii = yellow star; C. vinacea = red stars). All other entities shown on this map represent putatively undescribed species.

Fig. 13. Overview of the Clematis reticulata complex. Typical C. reticulata is represented by the pink circles with black border in the southeastern U.S. Coastal Plain. Orange circles in Alabama and adjacent states represents the previously described Clematis subreticulata, a species that probably needs to be resurrected. All other entities represent undescribed species.

Nearly all of the new species to be recognized are diagnosed by a syndrome of morphological features but no single species can be separated by any one feature. The various taxonomically important features in Clematis such as leaf dissection, inflorescence type, flower color, bract size and placement, sepal grooving, pubescence, etc., all are recombined in various ways such that oftentimes each major ecoregion has its own distinctive Clematis.

In the third part to this series I will provide photographs and tentative range maps of all of the members of the C. viorna and C. reticulata groups to help guide botanists in the eastern U.S. in interpreting variation of the eastern North American leatherflowers.