Dichelostemma capitatum ssp. pauciflorum, "few-flowered blue dicks"
Table of Contents
Number of Flowers per Head
Number of Pedicels Exceeding Bract Tips
Longest Pedicel Relative to Bract Length
Perianth Lobe Angle
Photographs in fig. 1 not added yet!!!!
Fig. 1. Left: , . Right: , Click on the pictures for larger versions.
There is much confusion about how to distinguish Dichelostemma capitatum ssp. pauciflorum from ssp. capitatum. To try to understand the sources of that confusion, I analyzed iNaturalist photographs of over 100 flowering plants at eight locations. Those locations spanned a distance of 1,000 miles (1,600 km), from the type locality of ssp. capitatum in Monterey Bay, California to the type locality of ssp. pauciflorum in Grant County, New Mexico.
I found two major sources of the confusion. First, there are major errors in floras for the range of variation of the two subspecies. Subspecies pauciflorum was found to have at least 3 to 9 flowers per head, not the "2 to 5" given in the floras. The majority of pedicels in its inflorescence do not exceed the bracts, despite the floras flatly declaring "pedicel longer than bracts". The bracts of ssp. capitatum are not "usually dark purple"; fewer than 50% of examined specimens have dark purple bracts. More often they are either a "light purple with darker purple streaks" or "white streaked purple".
Second, the morphology of the two subspecies has considerable overlap. As a result, some specimens cannot be clearly assigned to a subspecies morphologically, and subspecies determinations must be made using population studies examining on the order of ten plants in a population.
Subspecies capitatum appears to be restricted to the California Floristic Province (CA-FP) and immediately adjoining areas. Subspecies pauciflorum is found only in Arizona, New Mexico, and northern Mexico. The population in the Mojave National Preserve, southernmost Nevada, and east to Kingman, AZ appears to be intermediate between the two subspecies. More study is needed of that population to see if it might be an undescribed subspecies, or if it simply consists of plants intermediate between the two subspecies.
In the last several decades, Dichelostemma capitatum, the common blue dicks, has been shown to be distinct from other species of Dichelostemma in both morphology and molecular evolution, and is now called Dipterostemon capitatus in both the Jepson eFlora and iNaturalist.
While there is now some consensus about the scientific name of the species, there remains significant confusion about ssp. pauciflorum. None of the latest studies has shed any new light about how to distinguish ssp. pauciflorum from the rest of the species.
Stimulated by ongoing confusion at iNat about this subspecies, and in particular by this discussion, I studied some of the plethora of photographs at iNat to try to quantitatively understand the morphology of this subspecies compared to ssp. capitatum. iNat is a wonderful tool to study this subspecies, since it contains a whopping 16,340 observations of the species as a whole. Furthermore, all the characters used to distinguish this subspecies can be easily seen in photographs of the inflorescence taken from the side, as shown in Fig. 1.
D. capitatum is a very widespread species in the west, as shown in the SEINet map reproduced in Fig. 2.
Fig. 2. The geographic distribution of vouchers, augmented with a few observations, of D. capitatum, from SEINET as of 12 July 2021, with the subspecies indicated by the colors in the legend. The type locations for two subspecies are marked with arrows.
Fig. 2 shows that ssp. capitatum is mostly found in the California Floristic Province (CA-FP), with its type locality in Monterey in the heart of the CA-FP. Subspecies pauciflorum is mostly found to the east of the CA-FP, mostly in the higher elevations of the desert mountains of California, and the mountains of Arizona, New Mexico, and extreme northwest Mexico. Its type locality is at the extreme eastern end of the range of the species, in Grant County, New Mexico.
However, close examination of the map shows that there are a number of specimens determined as ssp. capitatum scattered throughout the area dominated by ssp. pauciflorum. There are also ~25 specimens determined as ssp. pauciflorum within the CA-FP; see the Calflora map.
I picked the eight locations shown in Fig. 3 to study characteristics that could be measured in iNat photographs.
Fig. 3. The eight locations selected to study Dichelostemma photographs, shown on a map of iNat observations. At each location, I sized the box so that I would retrieve about 20 to 30 iNat observations. Large boxes had low spatial densities of iNat observations.
Since my main goal was to understand the plants found in the Borrego Desert area, I picked locations mostly at the latitude of the Ranchita - Borrego Springs area. The Baker to Kingman area was used because there are no Dichelostemma plants found in the Colorado Desert, and because Jim Andre had reported that the Dichelostemma plants in that area were not clearly one of the varieties. Finally, the Monterey area was selected since it was the type locality for ssp. capitatum.
My initial goal was to measure all the characteristics that were said to be useful in separating these varieties: the number of flowers per head, the inflorescence bract color, the number of pedicels that exceeded the inflorescence bract length, and the angle of the perianth lobes from the axis of the flower. In addition, I planned to measure the bract length, pedicel maximum length, the flower tube and lobe lengths and widths, and the perianth width and widths, on photographs of the inflorescence from the side. Since the photographs had no scale, I divided all of those measurements by the bract length to get scaled numbers. My intent was to put all of those measurements into a PCA to see how well it could determine the subspecies.
Unfortunately, although some areas had a number of excellent photographs of the inflorescence from the side, other areas had essentially no such photographs, leaving me with too few observations in some areas to do a PCA.
Instead, I measured what I could on each photograph, and present plots below of what I could measure. The number of measurements of each characteristic ranged from a low 27 for the flower tube length, to a high of 130 for the number of flowers per head. The measurements were plentiful enough to give a good understanding of the range of measured characteristics for each subspecies.
Number of flowers per head
The latest treatment of these two subspecies, in the 2019 revision of the Jepson eFlora by Preston and Pires, separates them by "flowers 2 to 16(25)" for ssp. capitatum, vs. "flowers 2 to 5" for ssp. pauciflorum. However, the 1993 Flora of North America (FNA) treatment by Pires separated them by "flowers 6 to 15" for ssp. capitatum, and "2 to 5" for ssp. pauciflorum.
The number of flowers per head measured from a photograph should be considered a lower limit to the actual number of flowers per head in some cases, especially when there is a large number of flowers. Many heads contain small buds which can be hard to see, and heads with a large number of flowers can also have some flowers hidden behind others. Some measurements, of heads with few flowers that have multiple photographs from different angles, which does occur, are accurate. I noted in the data I took whether a measurement could be a lower limit or was an accurate count.
A plot of the measured number of flowers per head, vs. longitude, is given in Fig. 4. This and subsequent plots are mostly of each measured characteristic vs. longitude, since the eight study locations are all separated by longitude. Each locality is color coded as well.
Fig. 4. Individual measurements of the number of flowers per head plotted versus longitude, with the eight localities color coded..
All specimens from the easternmost three locations, all presumably ssp. pauciflorum (see further data below), have 3 to 9 flowers per head. Clearly, the "2 to 5" flowers per head for this subspecies is incorrect, and undoubtedly why some specimens from those areas have been determined as ssp. capitatum.
The locations to the west all have a larger maximum number of flowers per head. All are within the maximum of 25 flowers per head for ssp. capitatum except for this observation with at least 31 flowers (the topmost middle cluster). Those flowers appear to be part of a single head, from their symmetry, but it is possible this was a "doubled inflorescence". Unfortunately, there was no side view to see if it looked unusual.
The Baker to Kingman population, the orange diamonds, appears to be more closely aligned with the populations to the east, since the number of flowers per head is mostly 4 to 8. However, there are four individuals with 10 to 12 flowers per head, complicating the interpretation. If these are considered to be part of ssp. pauciflorum, then the number of flowers for ssp. pauciflorum occasionally extends up to 12 flowers per head. Or this population might be considered to have some intermediacy with ssp. capitatum.
In any case, the number of flowers per head is clearly not a very useful discrimination for an individual plant, although it does appear useful to characterize the population as a whole. I.e., if there are no individuals with more than ten flowers per head, it is likely this is a population of ssp. pauciflorum. Likewise, if there are a number of individuals with more than ten flowers, it is likely this is a population of ssp. capitatum.
Fig. 5 shows the median number of flowers per head at each location, which shows there is not much difference in that number between some areas of ssp. capitatum and areas of ssp. pauciflorum. This reiterates that there are many individual plants that cannot be determined to a subspecies based just on that number.
The two locations with a much higher median number of flowers per head are the Elfin Forest and Daley Ranch. Both of those locations had the fewest measurements, just five for the Elfin Forest and 11 for Daley Ranch, compared to 13 to 25 measurements for the other points. There is also a strong non-statistical effect at each locality, since many localities are dominated by a single observer at iNat. If that single observer only photographs the showiest flowers with the larger number of flowers per head, that would bias the number higher.
Fig. 5. The median number of flowers per head at each locality plotted vs. Longitude.
Fig. 6 shows a histogram of the number of flowers per head for all areas combined for each subspecies. For ssp. capitatum I used the westernmost four areas of Monterey, Elfin Forest, Daley Ranch, and Ranchita to Borrego Springs. For ssp. pauciflorum, I used Baker to Kingman, NW of Phoenix, East of Tucson, and the Grant County area. In order to produce a histogram not dominated by noise, I had to use bins consisting of three numbers of flowers per head. I.e., the point plotted at 3 flowers per head is the sum of the counts for 2, 3 and 4 flowers per head. Similarly, the point plotted at 6 flowers per head is the sum of the counts for 5, 6 and 7 flowers per head.
The histogram for Baker to Grant contained 69 measurements, and the histogram for Monterey to Borrego had 60 measurements. In order to make the histograms directly comparable, the Baker to Grant histogram was scaled by 60/69.
Fig. 6. Histogram of the number of flowers per head, using bins of three counts, for the westernmost four locations, presumably ssp. capitatum, and for the easternmost four locations, presumably ssp. pauciflorum.
There is clearly a difference in the histograms for the subspecies as a whole, but as mentioned above, the difference is not terribly useful except at the population level.
Both the 2019 Jepson eFlora treatment, and the 1993 Flora of North America treatment, separate the two subspecies by "bracts usually dark purple" for ssp. capitatum, and "bracts whitish or streaked purple" for ssp. pauciflorum. Fig. 7 shows the bract color for all iNat observations I examined that show the bracts.
Fig. 7. Plot of bract color on each specimen for which it could be measured vs. longitude. A value of zero is entirely white. A value of one is "white streaked purple". A value of two is "light purple streaked darker purple". A value of three is "dark purple". See also a version of this plot that has those labels in the plot.
The three easternmost locations all have bracts entirely white or white streaked purple, perfect ssp. pauciflorum according to the key.
The four westernmost locations don't fit the "bracts usually dark purple" of the key for ssp. capitatum. Even the ssp. capitatum type locality only has two out of six plants with dark purple bracts, with four plants having light purple bracts with darker purple streaks, and one plant having white bracts with purple streaks.
Just as with the number of flowers, the bract color is very good at a population level. The three easternmost locations are the only ones to have some plants with pure white bracts, and no plants with dark purple bracts or one that are light purple with darker streaks, signifying a population of ssp. pauciflorum. The four westernmost locations are the only ones to have no plants with pure white bracts, and at least one plant with dark purple bracts, signifying a population of ssp. capitatum.
The Baker to Kingman population appears intermediate between the other groups, but this time it aligns more closely with the ssp. capitatum groups, with only a single individual having the pure white bracts that only ssp. pauciflorum has.
Number of Pedicels Exceeding Bract Tips
Both the 2019 Jepson eFlora treatment, and the 1993 Flora of North America treatment, separate the two subspecies by "pedicel usually shorter than bracts" for ssp. capitatum, and "pedicel longer than bracts" for ssp. pauciflorum. Fig. 8 shows how many pedicels for each inflorescence that could be measured that were longer than bracts.
Fig. 8. Plot of the number of pedicels in each inflorescence that were longer than the bracts vs. longitude. Note that multiple inflorescences can plot at the same values for each locality, especially for the westernmost localities that are ssp. capitatum. For example, three points are seen for the orange diamonds in the middle of the plot, but they represent six inflorescences.
Pedicels elongate as the flower matures, so this number per inflorescence changes with time. Of course, such changes affect all locations, so comparisons between locations should be valid. Also, to try to minimize the effects of such changes, I did not record values if most of the flowers were in bud, or if most of the flowers were in fruit.
It is quite clear from Fig. 8 that the majority of pedicels in the inflorescence of ssp. pauciflorum do not exceed the bracts! In the 11 inflorescences I measured at the type locality of ssp. pauciflorum, only 11 pedicels out of 64 flowers exceeded the bracts. For example, this observation, with four finished flowers, two open flowers, one opening flower, and two buds, has no pedicels exceeding the bracts.
However, there is a clear trend that generally, but definitely not always, ssp. pauciflorum has a larger number of pedicels exceeding the bracts than does ssp. capitatum. The major exceptions occur in the Elfin Forest, at Daley Ranchhttps://www.inaturalist.org/observations/79867561
In this characteristic, the plants from Baker to Kingman align with ssp. capitatum.
Longest Pedicel Relative to Bract Length
To get an idea of how long the pedicels were relative to the bract length, I measured both lengths in photographs taken from the side where I could clearly see both parts. Fig. 9 shows the results.
Fig. 9. Plot of the longest pedicel length in an inflorescence divided by the longest bract length, vs. longitude.
There were no suitable photographs to use for this measurement in the westernmost two locations, so no values for them are shown.
The easternmost two locations had a number of specimens that had similar values to ones at other locations, but also had a number of specimens with significantly higher ratios. There were too few specimens measured to make meaningful comparisons for the other locations.
Perianth Lobe Angle
The perianth lobe angle, specifying how reflexed the lobes are relative to the axis of the flower, is not used in any key. The Jepson eflora states that the perianth lobes are "generally ascending" for the species as a whole, but gives an exception for ssp. pauciflorum as having "spreading" lobes. The FNA says that the lobes of ssp. pauciflorum are "widely spreading". Hoover (1940) even thought that the lobes of ssp. pauciflorum might be "rotate or reflexed at anthesis", but commented that this was difficult to see in vouchers. He wrote "Field study of var. pauciflorum and the apparently intermediate forms is greatly needed."
I made quick estimates from the photographs of the perianth lobe angle, using only photographs that were quite close to a side view. Photographs that had even a bit more of a frontal view made the lobes appear to have larger angles. The accuracy of my estimates is probably only something like 5 to 10°, due both to the unknown actual camera angle and my quick estimates.
A plot of the measured perianth lobe angle is presented in Fig. 10.
Fig. 10. Plot of the measured perianth lobe angle vs. longitude. A value of zero degrees means that the lobes were erect. A value of 45° means the lobes were ascending. A value of 90° means the lobes were spreading.
All measured lobes for both subspecies ranged from an angle of 10° to 45°, nearly dead erect to ascending. Not a single lobe was spreading at an angle remotely close to 90°.
There is a clear population trend for ssp. pauciflorum to have lobes making a higher angle from the vertical, with most specimens between 20 and 40°, and for ssp. capitatum to have lobes mostly making angles of mostly 10 and 20°. But the population full spread is almost the same for both subspecies.
In this characteristic, the plants from Baker to Kingman appear to be mixed in values between the two subspecies.
Most of the perianth measurements I made showed little difference between the two subspecies. Two of them, the perianth lobe length and the perianth diameter at its top showed some variation with location, shown in Figs. 11 and 12. For both of these, the two easternmost locations show a wider range of values compared to the other locations.
Fig. 11. Plot of the perianth lobe length divided by the longest bract length, vs. longitude.
Fig. 12. Plot of the perianth diameter at its top divided by the longest bract length, vs. longitude.
This study of a relatively small number of iNat observations was remarkably successful in understanding these two subspecies better, despite the considerable variation in how useful for measurements the photographs were among the observations. Of the 217 observations examined, 130 (60%) contained information on at least one characteristic, usually the number of flowers per inflorescence.
The three easternmost locations, from the Phoenix, Tucson, and Grant County areas, were all quite clearly ssp. pauciflorum, and showed that there were some very significant extensions to the range of characteristics of that subspecies.
The four westernmost locations, from Monterey Bay, Elfin Forest, Daley Ranch, and Ranchita to Borrego Springs, were all quite clearly ssp. capitatum, and also showed some significant extensions to the range of characteristics of that subspecies.
One location, the population in the Mojave National Preserve, southernmost Nevada, and east to Kingman, AZ appears to be intermediate between the two subspecies. It aligns with ssp. pauciflorum in its number of flowers per head. However, it aligns with ssp. capitatum in its number of pedicels that exceed the bracts, and the ratio of the longest pedicel to the bract length. It is intermediate between the two species in its bract color, and perianth lobe angle.
If I had been able to measure all those characteristics in a larger number of observations, a PCA might have revealed whether these specimens were consistent with being intermediates between the two subspecies, or whether they consist of an undescribed subspecies. More study is needed using better photographs taken from the side!
I thank Rupert Clayton, Carla Hoegen and Fred Melgert for stimulating this analysis, which has benefited from the findings of Carla and Fred from the field, and discussions with them and Rupert about the interpretation of their findings.
Copyright © 2021 by Tom Chester
Permission is freely granted to reproduce any or all of this page as long as credit is given to me at this source:
Comments and feedback: Tom Chester
Last update: 13 July 2021