Plants of Southern California: Stephanomeria diegensis


This page first presents the results of an analysis of the Stephanomeria diegensis population at Torrey Pines State Reserve on the Beach Trail / Broken Hill Trail Loop and the Guy Fleming Trail, as well as a population of six plants near the Visitor Center. All data and samples were collected on 29 November 2003.

The conclusion of this analysis is that all plants are identified as S. diegensis, being clearly separated from both S. virgata and S. exigua ssp. deanei.

Subsequently, a population at the Santa Rosa Plateau from the Mesa de Colorado was sampled and analyzed on 8 December 2003. Analysis shows that all those plants, previously identified as S. virgata by Lathrop and Thorne (1985), are also clearly S. diegensis, and essentially identical to the population at Torrey Pines.

Finally, the properties of the entire population so far identified as S. diegensis are presented.

See also a Photograph of Stephanomeria achenes, with the caption given here; and the geographical distribution of this species and closely-related species.

Data And Data Analysis: Torrey Pines

The gross morphology of all the plants was nearly identical, and was indistinguishable from that of Stephanomeria virgata, with ascending branches and nearly sessile flowers. Most plants were about 10 dm in height, with few shorter and some ranging up to 15 dm or so.

Detailed examination of the phyllaries and fruit revealed that there were two distinct populations. The population with non-glandular phyllaries had longer seeds and nearly always had five distinct grooves in the seeds, one on each surface. The population with apparently-glandular phyllaries had shorter seeds, and the number of grooves in each surface ranged from one to four. The distinction in the phyllaries and the grooves was noticed in the field. The distinction in seed length became apparent only in the later data analysis.

There were subtle differences in the gross morphology between apparently-glandular and non-glandular plants that led me to discover the first set of glandular plants, but those differences are hard to quantify. The glandular plants can perhaps best be described as being plants that seemed to resemble Stephanomeria exigua a bit, in having more spreading and more delicate branches. But no one would ever mistake these glandular plants for S. exigua, since the gross morphology remains that of S. virgata.

These populations appear to occur in distinct geographic regions. The non-glandular form is found in a contiguous area of at least 0.2 by 0.4 miles near the Visitor Center and the head of the Beach Trail. Specifically, it is found in the beginning 0.18 mile of the Beach Trail; in the 0.31 mile southeast and the 0.1 mile northwest of the head of the Beach Trail along the "Hiking and Biking Roadway", old US101. It is also found at the head of the Guy Fleming Trail. The glandular form is found in a different contiguous area beginning at mile 0.19 of the Beach Trail and extending to about mile 0.60 at nearly the bottom of the Beach Trail. Because I did not sample every single plant in each area, and because many plants had finished their life cycle and were dead, I cannot rule out the possibility that a few plants of one population appear amidst the many plants of the other population. However, I examined the phyllaries for well more than 50% of the blooming population in each area, including nearly 100% of the population on the Beach Trail itself, so there is no doubt that the vast majority of the plants in each area come from only one population.

I collected fruit and recorded the glandularness of the phyllaries from 11 locations. For the first five of these locations, I also recorded the number of flowers per head.

These 11 locations were not random. I sampled the first three locations of non-glandular plants and the first three locations of glandular plants. Subsequently, I collected data only when I thought there was a possibility of finding S. virgata, and hence selecting only non-glandular, taller plants. (I would have sampled any plants I thought might be S. exigua ssp. deanei, but none were found.) The last location was more randomly selected - the first specimen encountered on the Guy Fleming Trail.

In particular, this means that the ratio of non-glandular to glandular plants in my data is not representative of the actual distribution at Torrey Pines. I estimate that glandular plants are at least as frequent as non-glandular plants on the Beach Trail, despite the 3 glandular to 8 non-glandular ratio in my samples.

I subsequently measured 2 to 19 fruit from each location, recording how deciduous and plumose the pappus was, the number of grooves in each seed, and the length of each seed.

The pappus from every plant was readily deciduous. On the plants, it was difficult to pick up a fruit by its pappus; usually only the separated pappus was collected by pulling on it. Even when the connection of the pappus and the seed was maintained during collection, by the time I removed the samples from the film canisters in which they were stored, the pappus from every fruit had detached from the seed and collected into a roundish ball. Essentially none of the pappus remained connected to the seed.

The plumosity of the pappus was consistent from nearly every sample, ranging from ~75% to ~90% for the vast majority of the pappus, with a few samples at ~70% and a few at ~95%. (Each pappus for S. exigua and S. diegensis is plumose from its distal end to a point toward its base. Thus a pappus that is 90% plumose is plumose for the upper 90% of its length, and is a simple naked bristle in the bottom 10% of its length.)

The phyllaries fell into two distinct categories. The non-glandular phyllaries always were puberulent with soft white clean spreading hairs. In the field, the glandular phyllaries did not look like the typical glands of S. exigua ssp. deanei, with nice little balls of exudate at the end of spreading hairs. Instead, the hairs of these phyllaries were typically sticky, matted somewhat together or covered with dirt, with visible exudate usually sessile, infrequently stalked, on the phyllaries.

However, it is difficult to separate the normal stickiness of all parts of the Stephanomeria plants from any glandularity of the phyllaries. I did not examine any of these phyllaries from Torrey Pines under a microscope, only using a hand lens in the field. Examination under a microscope of the phyllaries from a similar population at the Santa Rosa Plateau (see below) does show some hairs with enlarged tips that look exactly like the ones from S. exigua ssp. deanei. The enlarged tips of both taxa do not change shape when poked with a dissecting needle. For brevity, I will simply call both of these glandular below, but the reader should keep in mind that this term may simply mean enlarged tips in this analysis.

The seed length for each plant typically was very consistent. For example, I measured 19 seeds from a plant at location #3, and every one was measured at 2.9 or 3.0 mm (3 were measured at 2.9 mm and 16 at 3.0 mm). No single plant produced fruit that varied by more than 0.2 mm in my measurement of its seed length.

However, the seed varied significantly from plant to plant, and correlates perfectly with glandularness:

(I rounded the mean seed length to the nearest 0.1 mm to produce the same accuracy in all samples. Some points have been dithered by 0.02 mm in seed length so that each location is seen separately in the plot.)

The glandular plants had significantly shorter seeds. Their individual seed lengths ranged from 2.0 to 2.4 mm, with the vast majority being 2.0 to 2.2 mm. In contrast, the non-glandular plants had individual seed lengths that ranged from 2.4 to 3.0 mm.

The number of seed grooves is also correlated with fruit length:

(The plants plotted with 2.5 grooves had 2-3 grooves per seed.)

The relationship is not as tight as that between glandularness and seed length, but it is significant. No glandular plant ever had seed with 5 clear grooves. Every plant with 5 clear grooves was non-glandular, with a longer seed length.

Classification: Torrey Pines

Can both sets of plants be classified as Stephanomeria diegensis? Let's review the classification and species present in this area.

None of these plants could be classified as S. exigua. They lack the gross morphology of that species, which normally consists of shorter plants, with more spreading and delicate branches, and flowers with much longer peduncles. Also, the phyllaries lack the obvious 100%-stalked glandularness of the local ssp. deanei.

Furthermore, none of these plants could be classified as S. virgata. Although the gross morphology is that of S. virgata, not a single seed was found that did not have at least one clear groove. Furthermore, none of the seeds looked like that of S. virgata, which has flat surfaces with noticeable ribs between them. Very few fruit had plumosity greater than 90%, whereas every S. virgata I've seen has plumosity of essentially 100%.

This leaves only S. diegensis, or a hybrid of S. diegensis with one of these two taxa, as the only possibilities.

First, consider the non-glandular plants with seeds that have five clear grooves. They key unambiguously to S. diegensis using couplet 12 in the Jepson Manual key, since all the other species in branch 12' are out of range except for S. diegensis.

That assignment is not without discrepancies, since the fruit length, the extent to which the outer phyllaries are reflexed, and the number of flowers per head is different from the JM treatment.

The mean fruit length for these plants is 2.7 - 3.0 mm, much longer than the 1.8 - 2.4 mm in the JM key and description. I did not do an extensive survey of the outer phyllaries, but the ones I observed had their tips spreading at about 45°, far from the reflexed position in the JM key and description. Finally, I observe the number of flowers per heads of these plants to be 5 - 10, not the 6-8 in the JM description. Interestingly, Gottlieb (1971), in the definition paper for S. diegensis, says that the average number of flowers per head is 11-13, far different from what is in the JM.

Ordinarily, I would reject any identification with such extensive and marked discrepancies, which would bring up the possibility of hybridization. However, I discount the disagreement for the phyllaries and the number of flowers per head since I have seen similar disagreements for S. exigua ssp. deanei and S. virgata, as well the discrepancy of the JM from Gottlieb.

Hence I consider only the fruit length to be a significant discrepancy. If that discrepancy is due to hybridization, the only possibility for the fruit length would be S. virgata, which has fruits of length 2.2-3.6 mm. That might then also explain why some of the plants have fewer than 5 clear grooves. It would remain unexplained why no other feature of S. virgata shows up in these plants, such as height and degree of plumosity for the pappus. Furthermore, one would have to explain the absence of any S. diegensis from its type locality!

So I think it is likely that these plants are pure S. diegensis, although I cannot discount the possibility that these plants are hybrids of S. diegensis with S. virgata, which would imply that pure S. diegensis is not found in these locations at Torrey Pines.

Second, consider the glandular plants, all of which have seeds that have less than five clear grooves. The clear hybridization candidate is S. exigua ssp. deanei, which would provide the glands. These plants do have fruit lengths that precisely fit the ranges given in the JM for both S. diegensis or S. exigua ssp. deanei. However, any such hybridization could not explain the fewer than five grooves, since S. exigua ssp. deanei also has five grooves. Furthermore, it would remain unexplained why no other feature of S. exigua ssp. deanei shows up in these plants, such as height and degree of plumosity for the pappus.

It seems more likely to me that S. diegensis simply can exhibit some glandularness at times. Since it is conjectured that the origin of S. diegensis is from a hybridization event of S. exigua ssp. deanei and S. virgata, it does not seem surprising to me that glandularity might be a variable trait of S. diegensis.

Furthermore, such an origin makes it not unexpected that the number of grooves would vary between the zero of S. virgata and the five of S. exigua ssp. deanei, nor that the length of the seeds would vary between the 2.1-2.4 mm of S. exigua ssp. deanei to the 2.2-3.6 mm of S. virgata.

The simplest hypothesis is that every plant I surveyed was a valid representative of the S. diegensis population, whose properties are simply more variable than that given in the JM. It is also not terribly surprising that a relatively new taxon would be less well-described than older taxa.

The following plot shows that all of these plants separate cleanly from both S. virgata and S. exigua var. deanei based on just the mean pappus plumosity (the average of the minimum and maximum plumosity found in individual samples) and number of grooves in the seed:

The specimens with glandular phyllaries are also indicated in the plot. The data for the other two species come from S. exigua ssp. deanei and S. virgata. The data points for S. virgata have been dithered to show all the points. In all cases, what is plotted above is the mean plumosity determined from all fruit collected from a plant or set of plants at a given location.

Of course, the habit of the inflorescence also is unambiguous in separating S. exigua var. deanei and S. diegensis.

Histogram Of The Number Of Flowers Per Head: Torrey Pines

The following histogram gives the distribution of the number of flowers per head for the non-glandular population, with 5 clear grooves, from the first three locations, along with propagated sqrt(N) one-sigma error bars:

Glandular specimens were excluded in the above plot, for those that wish only to accept non-glandular plants as pure S. diegensis. The data in the plot comes from a total of 46 heads.

The histogram below gives the same distribution, but now including data from three glandular plants observed on 11/28/03, as well as data from all plants observed on 2/15/02:

This plot used a total of 68 heads, including the 46 from the previous plot.

These histograms are identical to within the error bars, and both are consistent with a smooth, single-peaked curve within their error bars. It is also possible that the true histogram actually has two peaks, one close to 6-7 flowers per head and another close to 9-10 flowers per head. Such a distribution is also consistent with my histogram within the error bars. However, a distribution with two peaks is ruled out when the S. diegensis specimens from the Santa Rosa Plateau are included in a similar plot shown below.

Data and Data Analysis: Santa Rosa Plateau

I collected samples from the Santa Rosa Plateau (henceforth SRP) on 8 December 2003.

All specimens sampled near the Visitor Center, from the first ~0.30 miles of the Wiashal Trail and from the 0.70 mile portion of the Granite Loop Trail south of Waterline Road, were clearly S. exigua ssp. deanei, and will not be discussed further here.

I collected eight additional samples of a separate population about 2.5 miles south of the Visitor Center, from the Mesa de Colorado along the Vernal Pool, S. Los Santos and S. Trans-Preserve Trails. The morphology of this population was clearly different from S. exigua, and resembled that of S. virgata and S. diegensis. The eight samples were spaced fairly uniformly along a distance of ~1.0 mile north-south and ~1.0 mile east-west. The entire population was well over one hundred plants, and probably a minimum of several hundred.

Interestingly, Stephanomeria was present only on the basalt on the top of the Mesa and was not found elsewhere on these trails. These combined trails were roughly one-third on top of the Mesa and two-thirds below the Mesa. Fairly dense populations occurred in places on the Mesa immediately next to the edge, but none were found in the granitic soil just below the edge.

The only explanation I have found for this curious situation is that oak trees, Quercus agrifolia and Q. engelmannii, tend to form a fairly closed canopy just below the edge of the Mesas. It is possible that this produces a shady barrier that prevents the Stephanomeria population from extending into the grasslands below the Mesa.

I have plotted this population from the SRP on top of the plot presented earlier which separates the three Stephanomeria species based on pappus plumosity and the number of grooves in the seeds:

Many of the SRP data points fall directly on top of Torrey Pines data points, and they all fall cleanly in the S. diegensis region of the plot. The only difference between the two sets of data is a very small tendency for the SRP pappus to be slightly more plumose. However, the difference is not statistically significant in these data.

Properties of the S. diegensis population

Discrimination from the other Stephanomeria species. The following plot summarizes the separation of the three Stephanomeria species. It is the same as the previous plot, but now combines the Torrey Pines and SRP populations of S. diegensis instead of showing them with different symbols.

The population distribution of the number of flowers per head. I collected data on the number of flowers per head for the SRP population on 8 December 2003, and previously on 8 October, 10 November and 17 November 2001. Those datasets are shown separately below, along with the histogram from Torrey Pines.

The tremendous variability in different datasets is immediately apparent. I speculate that this variability, along with small-number statistics, is the cause of the discrepancy between the claim of 6-8 flowers per head in the JM and Gottlieb's claim of averaging 11-13 flowers per head.

(Small-number statistics applies whenever one does not sample a population sufficiently to reveal unambiguously the true properties of the population. When one is working with a small number of samples, random data can produce seemingly meaningful results at times. In this case, the Jepson Manual author may have seen a set of herbarium specimens that mostly had 6-8 flowers per head, and Gottlieb may have seen another set that mostly had 11-13 flowers per head. Both may thus have thought they had a good handle on the intrinsic distribution of the number of flowers per head when they actually did not.

Note that both authors may simply have had bad luck, and not bad judgment. It is sometimes very difficult to ascertain whether one has sampled the full range of variation for a species, and there are no guaranteed methods of knowing when you have done so.)

These datasets are combined in the following histogram:

This histogram contains data from 165 heads.

Again, note the variability seen in individual populations in the previous plot. One has to combine many different populations in order to observe the population-average histogram shown here.

The distribution of the number of flowers per head for individual plants. The maximum and minimum number of flowers observed for individual plants are shown in the next plot:

There were fewer observations where the data for individual plants was kept separate, so this plot does not contain as much data as the previous plot. Points were dithered about their true position in this plot so that all the individual data points could be seen.

Although individual plants appear to be somewhat more restricted in their number of flowers, this may be due partly to small-number statistics. That is, a typical plant often has only 1-4 flowers in bloom at once for observation. Sampling the previous histogram with only 1-4 flowers per plant could very well result in a plot similar to the data shown above.

To address this possibility, I simulated producing the above plot by chance. I used the population distribution for the number of flowers per head, and randomly drew samples from that histogram for each head for each simulated plant. I then analyzed the simulated plant for its minimum and maximum number of flowers per head over all the simulated heads.

The following simulation for two heads per plant, for 36 artificial plants, which is the same number of plants as in the previous plot, produces a plot very similar to the actual data above:

Except for a few points, the bulk of the data are very similar. The exceptions go in both directions: the simulation produces both a single more-correlated point (a minimum of 13 and a maximum of 13) and a single less-correlated point (a minimum of 5 and a maximum of 13).

Thus considerably more data must be gathered to see if individual plants are more restricted in the range of number of flowers per head than what would be expected from random draws from the population distribution.

To show how challenging collecting more data actually is, I sampled 17 heads from one enormous plant at the SRP on 8 December 2003, which had 5, 5, 3, 3 and 1 heads with 8, 9, 10, 11 and 12 flowers, respectively. I then simulated picking 17 heads in the computer at random from the population distribution. The second simulation produced a range of 7 to 11 flowers per head, very similar to the 8-12 flowers per head observed for this plant. The first and third simulation both produced a range of 5-13 flowers per head, so perhaps this is weak evidence that individual plants have a slightly more limited range in their number of flowers per head. The reader can see that sampling even 17 heads from a single plant does not definitively indicate the range in the number of flowers per head for a single plant, when the population has such a wide intrinsic range in the number of flowers per head. Data sampling all heads from a hundred plants or so are required to answer this question definitively.

Properties of the outer phyllaries. I noted the properties of the outer phyllaries at the SRP on 8 December 2003. In all cases, just as observed for S. virgata at Agua Tibia Mountain, the outer phyllaries ranged from almost entirely appressed to have just their tips ascending to spreading. No phyllary was ever observed to have any portion reflexed, despite the Jepson Manual key and description for S. diegensis.

Height. Although the Jepson Manual says that the height of S. diegensis is 10-20 dm, the heights at the SRP on 8 December 2003 ranged from 10-30 dm. The heights at Torrey Pines on 28 November 2003 were significantly shorter, ranging from 10-20 dm. However, many species are shorter when growing immediately next to the coast, so I do not consider this difference significant.

Stem branching. The Jepson Manual also says STS: much-branched. While that is true generally, on 8 December 2003, and on 18 September 2001, I observed a number of plants at the SRP with essentially no branches, just a very long single stem with a few minor short branches. I sampled one of these on 8 December 2003, and the pappus and seed were no different from the other S. diegensis.

Clustering of flowers. The JM says heads solitary or clustered on short, stiffly-spreading branches. I observed few, if any, solitary heads, but I did not make a specific survey for them. The heads were almost entirely in clusters averaging ~4-5, with up to ~9, flowers, with each flower having its own peduncle attached directly to the node, without any short branches. These flowers were either on the main stem at its nodes, or on the typically very-long ascending branches on their nodes. The length of the peduncles ranged from 0 to ~4 mm in a single branch of a single plant that I examined.

In other words, the plants I observed bear no resemblance to the inflorescence structure described in the key for this taxon, nor to the JM drawing of S. virgata ssp. pleurocarpa, which has the same inflorescence description.

See Comments on the Jepson Manual and A Flora of Southern California by Munz: Stephanomeria for a revised key to this species.

I thank James Dillane for his help in the 11/28/03 survey, and Dieter Wilken for information from the Gottlieb reference.

Go to:

Copyright © 2003 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: 19 December 2003