Plants of Southern California: Pilostyles
Table of Contents
Introduction
Distribution
Infestation Properties
Introduction
Pilostyles thurberi, Thurber's stemsucker, is called Southern California's Most Unusual Wildflower by Wayne Armstrong, and it would be hard to dispute that claim. P. thurberi is a parasite on a common shrub in Sonoran Desert washes, Psorothamnus emoryi (=Dalea emoryi), commonly called Emory's indigo-bush or dyeweed. Like many parasites, there is no visual indication that it is present on a host until it blooms.
P. thurberi is only found in southwestern North America, from Mexico to California to Nevada and Texas. In California, it is found only in a small range near the Salton Sea, even though its host plant ranges over a much larger territory.
Job Kuijt, author of The Biology Of Parasitic Flowering Plants, points out that the genus Pilostyles has one of strangest geographic distributions of any plant species. The genus Pilostyles is also found in the Near East, and in Ethiopia and Australia. (See distribution map.) It is very puzzling how a genus could end up with a presence only in those widely-separated places. The distribution becomes only a bit more understandable if Pilostyles is not actually separate from Apodanthes.
In the rest of this article, I will use the name Pilostyles to refer only to P. thurberi, and not to the genus.
The flower of Pilostyles is so small that most people would walk right by a plant with dozens of its blooms and never notice it. This is a bit surprising, since it would be very hard to miss the bloom of a close relative of Pilostyles, Rafflesia arnoldii, commonly called Corpse Flower or Stinking Corpse Lily, which has the world's biggest individual flower. The Corpse Flower also has, shall we say, a distinctive smell, which gives rise to the common name. (Although both species have the same common name, don't confuse Rafflesia arnoldii with the better-known Amorphophallus titanum, which is cultivated and stars in the press whenever it is about to bloom!)
See Wayne Armstrong's delightful article on Pilostyles, with many fantastic pictures both of Pilostyles and its larger relative.
Beauchamp gives a bloom time of January - May, although Munz gives just January; Wayne observed plants in bloom at Christmas some years ago; and Wayne was told of those plants blooming in November in a prior year.
In late February 2005, Wayne Armstrong and I went Fonts Point Wash in Anza-Borrego Desert State Park and readily located a handful of Psorothamnus emoryi plants with old Pilostyles blooms on them, and I GPS'd their locations. The old blooms looked like they had been dead a long time, possibly more than just in the last year, so we suspected we had at least missed the bloom in late 2004 or early 2005.
In late 2005, I made it a priority to try to catch Pilostyles in bloom, and began checking the plants in October. On 19 December 2005, I caught the first plant in bloom; see Paula Knoll's stunning pictures of this plant.
The pictures by Wayne and Paula are quite misleading, as is often the case when good photographers take pictures of minute flowers. To the naked eye, the open flower of Pilostyles is barely noticeable as a flower and in fact, hard to distinguish from the dead, dried old flowers. On 19 December 2005, when I finally came across a plant in bloom, it took me a moment or two to finally realize I was seeing fresh flowers. Yet Wayne and Paula are able to take that almost-nothing flower and make it look like a movie star!
The following sequence of pictures will give the reader an idea of what it looks like to the eye:
Paula Knoll next to the host plant, Psorothamnus emoryi, dyeweed:
You can see that a lot of people might even walk by the host plant when it is in bloom, since its blooms are somewhat less than a spectacular show!
View of the bloom of the host plant, Psorothamnus emoryi, dyeweed, approximately as seen by the human eye, with Pilostyles blooms on the stem of the host plant, at lower right:
Magnified bloom of the Pilostyles, appearing about two times larger than would be seen by the unassisted human eye:
Compare the above picture, already magnified about two times over unassisted human sight, with Paula Knoll's pictures of this same plant!
After seeing its bloom, I of course became curious about this species. It turns out almost nothing is known about it!
In particular:
- As you can see from above, not even its bloom season is known very well.
- The detailed distribution of Pilostyles is unknown, other than the locations mentioned above. When I first read those locations, I immediately thought that Pilostyles might be restricted to Colorado River sediments, but this is apparently not known one way or the other.
- Nothing is known about the rate of infestation of host plants.
- Nothing is known about its pollination, which is bound to be interesting since it appears to be the case that each plant of Pilostyles produces either only male or female flowers. (Each flower is unisexual, but I couldn't find a reference that stated whether each plant only produced flowers of the same sex. So far, in a very limited sample, Wayne and I have observed only flowers of a single sex on a given plant. Job Kuijt informs me that the vast majority of species in Rafflesiaceae are strictly dioecious.)
Job Kuijt reports that he has seen two different sexes on a single host plant from two separate Pilostyles plants growing on that host plant.
- Nothing is known about the dispersal of the seeds.
- Nothing is known about germination of the seeds, not even whether the seeds germinate in the ground, like Cuscuta, dodder; or whether the seeds germinate on a branch of a host plant, like mistletoes.
Wayne Armstrong and Job Kuijt provided some of the above interesting information on what is known and what is not known about this species.
Stimulated by the above, on 28 December 2005 I decided to do a little fieldwork to find out a bit more about this species. I was especially interested in the distribution, since I found it quite bizarre that somehow the parasite might care about what soil its host plant is growing in!
I subsequently collected more data on 11 and 15 December 2007.
The following data and analysis come from that fieldwork, except for the plot of previously-known locations.
Distribution of Pilostyles
Distribution From 28 December 2005 Fieldwork
To begin to understand the distribution of Pilostyles, I surveyed an area of Fonts Point Wash immediately south of S22, as a baseline, finding a number of plants of Pilostyles. I then moved west to try to find the westernmost area of occurrence. The next wash to the west is Inspiration Wash, which also drains the Badlands, which are made of Colorado River sediments. To a first approximation, Inspiration Wash has the same rate of occurrence of Pilostyles as Fonts Point Wash.
The next wash to the west, and the first wash that does not drain the Badlands, is Coyote Creek Wash in Borrego Valley, which is different from the Coyote Creek Wash below Borrego Valley. I walked around that wash where it intersects Henderson Canyon Road, but that wash has been totally trashed, probably by farming, and has no Psorothamnus emoryi.
I (and others) had previously surveyed Coyote Creek Wash upstream of that location, near the end of the pavement on Di Giorgio Road, and found no Pilostyles. But to make sure that I just hadn't been able to recognize the Pilostyles there previously, I stopped there and resurveyed 100 plants of Psorothamnus emoryi, finding no Pilostyles.
Just to make sure, I continued driving up Coyote Creek Road, and surveyed a small wash just to the north of the first Coyote Creek Wash crossing. I only had time before sunset to find five plants of Psorothamnus emoryi, none of which had Pilostyles on them.
This distribution is consistent with the northwestern edge of the Pilostyles distribution being found within the Colorado River sediments.
In fact, within each of the two washes in Colorado River sediments, the detailed distribution shows supporting evidence for confinement of Pilostyles to those sediments.
The next two plots show the location of every plant of Psorothamnus emoryi in the area I mapped in those two washes, with each plant that showed flowers of Pilostyles, dried or fresh, marked separately:
 | 
|
A scale, showing a distance of 100 feet (30 m) is given at the bottom of each map. I've slightly enlarged the points corresponding to plants with Pilostyles to make them more visible. Even so, the two northernmost locations at Fonts Point Wash are hard to see amidst the sea of plants without Pilostyles.
In both plots, the main wash, with essentially no plant growth at all, is the blank area trending north-northwest to south-southeast. The Psorothamnus emoryi plants plotted are mostly found on the immediate edges of the washes.
The Fonts Point Wash survey was essentially confined to the area marked by displayed symbols in the plot, since Psorothamnus emoryi is ubiquitous there. At Inspiration Wash, I also surveyed an area to the north of the area shown in the plot. That portion was about three times as large as the area shown, and is not displayed since it contained essentially no P. emoryi plants. That area is almost entirely an Ambrosia dumosa forest.
At Fonts Point Wash, on the east bank, the P. emoryi population dies out to the south since the drainage is incised below the east bank to the south. The drainage is incised below the west bank to the north of the plotted points.
It is striking that the number of host plants with Pilostyles is higher to the south, which is toward the center of the Badlands.
The next plot quantifies the percentage of host plants with Pilostyles, using bins of width 0.0005 degrees of latitude which corresponds to roughly 200 feet of distance. The plot shows that near the location of highway S22, the rate of infection of Psorothamnus emoryi decreases rapidly to the north. Highway S22 is located, not by coincidence (few highways traverse badland areas!), at essentially the boundary where the Colorado River sediments have been stripped away by erosion.
The rate of infection is 25-30% in areas to the south of S22 where there are still sediments from the Colorado River, and the rate decreases rapidly beyond the end of those sediments.
Distribution From 11 and 15 December 2007 Fieldwork
On 11 December 2007, I surveyed an additional 1.3 miles south of the end of the 28 December 2005 survey in Fonts Point Wash, with the goal of GPS'ing only plants with Pilostyles.
Much to my surprise, the host plant, Psorothamnus emoryi, was nearly absent in this section. I saw perhaps only ten total plants in the 1.2 miles further up-wash I covered. Although none of those ten were infected, the rate of infection is still consistent with the 2.7 ± 1.6 plants expected from the rate near S22 at the 95% confidence level (two times the 1.6 standard deviation).
I suspect the sharp boundary between areas with and without P. emoryi is because that species only likes some of the rock formations in the Badlands. Since the formations are tilted significantly, one encounters new layers within a short distance. In fact, a significant portion of this area was almost totally barren, nearly absent of any species of plants, due to a such high salt content in the soil that there were numerous salt crusts on the surface of the ground.
On 15 December 2007, Wayne Armstrong and I continued the Pilostyles survey, this time in the neighborhood of Fonts Point, four miles south. We surveyed a total linear distance of 2.0 miles, 0.9 miles from Fonts Point down the road to just east of Inspiration Point, then 0.7 miles up a wash that ended at the rim 0.4 miles east of Fonts Point, and back to Fonts Point.
We found only a handful of Psorothamnus emoryi, about half with Pilostyles, all in a single location at Fonts Point. The absence of Psorothamnus emoryi below Fonts Point cannot be explained by a different rock formation, since 0.9 miles of our survey was in the same formation, Fonts Point Sandstone, and we saw the same species throughout that 0.9 miles.
Elevation appears to be the other variable controlling the Psorothamnus emoryi distribution. Although the Jepson Manual gives the habitat for this species as desert flats, washes, dunes; < 700 m (2300 feet), Munz 1974 gives it as dry open places below 1000 feet. From these two sources, one suspects that this species only really likes to live at elevations below 1000 feet, and that there are only scattered populations above that elevation. Since the elevation of Fonts Point is 1294 feet, and the elevation of the part of the wash near S22 is 700 feet, this could explain the general absence of plants in the upper part of the wash.
A quick examination of vouchers finds this suspicion seems to be the case. The following plot gives a histogram of vouchers of Psorothamnus emoryi that have reported elevations:
Most vouchers are found near sea level, with almost linearly-declining numbers above that elevation; only 13% of the vouchers are found above 1000 feet. Furthermore, most of those higher-elevation vouchers are from the Mojave Desert.
Confirming this conclusion, the nearby Deep Canyon, Santa Rosa Mountains Flora gives the elevation range for this species as 0 to 700 feet.
So now the distribution makes some sense. The species simply likes to live mostly below ~800 feet here, with only scattered populations above that elevation. Presumably this species is quite sensitive to frost, and locations above ~800 feet are usually too cold for it unless they are in frost-protected locations.
Distribution From Other Sources
Munz just gives Imperial County for its distribution; the Flora of San Diego County (Beauchamp 1986) gives Coyote Creek, below Borrego Valley; Fonts Point Wash; below the Carrizo Overlook; ca. 4.5 miles west of the Imperial County line on Highway S-2. Craig Reiser gives a number of locations from vouchers and sightings in his book Rare Plants of San Diego County. Larry Hendrickson kindly shared a very interesting location where he had observed it, above Clark Lake.
The locations are plotted in the map below:
Note that nearly all of the locations are imprecise; I have simply arbitrarily plotted a location consistent with each location.
All of the locations above roughly the latitude of the town of Brawley are in the Colorado River sediments, as located on the Santa Ana Sheet in the 1:250,000 scale Geologic Map of California. Larry Hendrickson's observation, the most northwestern point plotted, in especially interesting since my data showed that Pilostyles appeared to vanish at the northern boundary of the Badlands, which is where the Colorado River sediments have been eroded away. (The location shown just above the g in Borrego Springs, the nearest location to Larry's location, is in the Badlands.) Larry's observation shows that the distribution of Pilostyles picks up again to the northwest of the Badlands, precisely where the Colorado River sediments pick up again. This bolsters the hypothesis that Pilostyles is somehow confined to those sediments, but is still a long way from proving that this is the case.
I have not yet consulted the San Diego - El Centro Sheet, in order to compare the rest of the locations against the geology. However, purely from the topographic map, it seems highly likely that all of those points are also in Colorado River sediments, except for one. The single location that may not be in Colorado River sediments is a voucher "in the Jacumba Mountains east of Carrizo Canyon".
This location is of course quite imprecise; it may well be on Colorado River sediments, but it is not obvious or perhaps even likely from the location that it is.
There is a voucher location of Psorothamnus emoryi, the host plant, plotted in Sonoran Desert Plants (Turner, Bowers and Burgess 1995) at approximately the location of the Jacumba Mountains, but at an elevation of 250 feet. It is likely this is the same voucher, since Turner et al plotted vouchers from local herbaria in addition to their own observations. I would not call any location at 250 feet as being in the Jacumba Mountains, since it is fairly clearly away from the mountains. However, although 250 feet is just 5 miles from the eastern edge of the Jacumba Mountains, that elevation is definitely in Imperial County. Since that voucher was apparently from San Diego County, this appears to be a bit of a mystery. However, James Dillane points out that the voucher might have been collected before Imperial County was split from San Diego County. I'll check this voucher in the future.
Infestation Properties
In the following, I use the phrase "Pilostyles flower" to include both dead persistent flowers from previous year or fresh blooms from this year.
I use the single plant in bloom on 12/28/05 to deduce the properties of the Pilostyles bloom from year to year. This Psorothamnus emoryi host plant has about 10 cm of growth above the previous year's Pilostyles flowers, which I interpret as this year's growth in the two months since the 2.0 inches of October rainfall there. This year's new blooms are developing on that 10 cm of new growth.
The previous year's Pilostyles flowers were fully developed at the base of the stretch of stem in which they appear, but near the tip appear to have aborted as buds. (This progression can be clearly seen in the middle picture above.) This is consistent with those buds appearing at the very tip of last year's growth, and became aborted due to lack of nutrients. The lack of nutrients could be either because the stem at that point couldn't support a full Pilostyles flower, or because the host plant was shutting down due to lack of water at the end of the rainfall season that year.
There is at most a small separation of perhaps a few cm between the last of the previous year's flowers and the beginning of this year's flowers. I interpret this to mean that a continuing Pilostyles infection of the host plant will continue to produce flowers along a more or less continuous section of stem.
Thus if I see a host plant that has a lot of growth above the last Pilostyles flower, I conclude that Pilostyles is no longer alive within that host plant.
Studying the 32 Psorothamnus emoryi specimens with evidence of Pilostyles flowers gave me the clear impression that the Pilostyles infestation is short-lived on the host plant. This could result either from the Pilostyles plant itself being short-lived, or because the host plant eventually succeeds in eliminating the Pilostyles infestation.
I deduced this from the following observations:
- I found a few Psorothamnus emoryi plants with a total of only a very few flowers, all of which were well below the ends of the branches. The infection in these cases was clearly weak, and the Pilostyles plant bloomed only pathetically before dying. (This is some evidence that the host plant kills the Pilostyles infection.)
- I found no specimens with an interrupted Pilostyles bloom. An example of an interrupted bloom would be finding Pilostyles blooms along 10 cm of stem, then a section of 10 cm of stem with no Pilostyles blooms, and then finding another stretch of Pilostyles blooms. Instead, every specimen had only a single stretch of Pilostyles blooms.
- I found ten host plants that clearly had too much growth above the last Pilostyles flower for the Pilostyles to still be alive on those plants.
- The lengths of stems with Pilostyles blooms were in all cases consistent either with the blooms from a single year, or from two separate years. I found no cases where the length of blooms was consistent with having blooms from three or more separate years.
Although the above observation was a subjective one from the field, it was striking that there were no large host plants observed that had Pilostyles flowers present along a long stretch of branches. One might quibble about whether some plants exist that have blooms from three years or not, but it is extremely unlikely there are any plants with blooms from more than three years. Furthermore, this observation is consistent with the other observations given above.
I had intended to quantitatively measure the length of stem over which I could see evidence of Pilostyles blooms, but this proved to be too time-consuming on 12/28/05 to carry out for every plant while I was doing the geographic distribution as well. I may return to quantitatively investigate this in the future.
As implied above, for almost every surveyed Psorothamnus emoryi that had evidence of an infestation of Pilostyles, I estimated how many years of infestation were present. (I simply forgot to write down this estimate for two specimens.)
The single plant in bloom on 12/28/05 has a separation of ~5 cm (2 inches) between the last of the previous year's flowers and the beginning of this year's flowers. As a result, I had hoped to observe similar gaps between the bloom from one year and the bloom from the next. However, I found no such clear evidence in the field of a break between the bloom from one year and the bloom from the next on plants with only blooms from previous years.
In practice, it is fairly easy to estimate the number of years, because the length of stem with old blooms was either clearly consistent with bloom from a single year, clearly consistent with bloom from two separate years, or consistent with blooms from either one or two years. (Quantitative measurements might be able to distinguish between one or two years, or variability in the length of bloom along a stem might make it fundamentally impossible to distinguish the number of years of bloom in some cases.)
In addition, I noted how much new growth on the P. emoryi appeared after the last Pilostyles flower on the stem. If there was only one year's growth, similar to what was observed on the single plant in bloom, I considered the Pilostyles plants to still be present on that host. If there was clearly two or more years of growth on the host plant above the last Pilostyles flower, I considered the Pilostyles plant to be dead on that host.
The following table gives the results of this field survey:
| Min-Max # Years of Pilostyles | Total specimens | Pilostyles Still Alive | |||
|---|---|---|---|---|---|
| Yes | Maybe | No | Not Estimated | ||
| 1-1 | 20 | 14 | 6 | ||
| 1-2 | 6 | 2 | 4 | ||
| 2-2 | 5 | 1 | 3 | 1 | |
| ? | 1 | 1 | |||
| All | 32 | 1 | 19 | 10 | 2 |
Pilostyles was clearly no long present in one third of the specimens (10 out of 32).
No firm deduction can be reached on the typical number of years of Pilostyles blooms until I find out how many of these plants will bloom this year or not. That is, it is still possible that all of the 19 plants marked Maybe in the above table could still blooms this year, and thus have two years worth of Pilostyles blooms by the end of this year.
I plan to return later this year to follow this up.
Other observations from the 12/28/05 survey: Amount of infection on the plants: I observed some plants that had only one branch infected, but most of the time, nearly all branches of a plant are infected. For example, there was a cluster of ~9 very old Psorothamnus emoryi plants in the northwest section of my Fonts Point Wash survey. None of the plants had any infection except for a single plant, which had a single branch, out of 20, that had flowers for 1-2 years, at the very base of the plant.
Although it is tempting to interpret plants like this as evidence for seeds germinating on a stem, infecting only that stem before being killed, that observation is also consistent with an infestation from the root, which was only strong enough to flower on one branch before being killed.
There is very strong evidence that infestations can happen "from the ground up", since there were a number of plants with very heavy infestation at the very base of all stems of the plant.
Preliminary conclusions from detailed measurements made on 11 January 2006. The following conclusions are from an email report of a quick analysis done on 11 January 2006 immediately after making detailed measurements on a number of the host plants with Pilostyles in Fonts Point Wash and Inspiration Wash near S22. I have not had time to do a full analysis of that data, but thought it would be useful to others to have these preliminary conclusions online.
The email is reproduced exactly as sent, in order to emphasize that this has not been properly analyzed in detail. In particular, since the preliminary conclusions have not been subjected to thorough analysis, they may be wrong! I make no guarantees here; use at your own risk.
wayne and i then settled in for a long afternoon's work on the plants with pilostyles, both here and at inspiration wash. i concentrated on measuring the psorothamnus emoryi host as well as the extent of the visible pilostyles on each host plant. wayne concentrated on trying to find what animals were in the vicinity of each plant, to look for potential pollinators. we both tried to assess the condition of the pilostyles infection on each plant.
virtually every p.e. plant has captured a significant mound of sand around its base. wayne noticed right away that every plant seemed to have big holes in the sand leading under it, which wayne suspected were due to kangaroo rats. this certainly made sense, since everyone checking out the pilostyles has postholed into tunnels near those plants!
wayne found a very large (about a foot long!) nail / screw discarded along the road, which came in handy to poke into those holes to see what came out. to our surprise, ANTS came pouring out.
wayne quickly noticed that the ants were discarding p.e. perianth parts at the edge of their wastepiles, and the light dawned. ANTS are probably the pollinators of pilostyles! they have to crawl over the pilostyles flowers along the branches of the p.e. TWICE to get to each terminal p.e. flower. it would be hard for them NOT to be pollinating the pilostyles flowers on their travels.
in the middle of the range for pilostyles, 1/4 of all the p.e. plants are infected, so even if each p.e. plant with pilostyles only has a single sex of pilostyles, it seems almost guaranteed that some ants will end up visiting pilostyles flowers of both sexes.
wayne caught a great pix of an ant dragging a p.e. calyx, which is very readily identified by the orange glands. see:
http://waynesword.palomar.edu/ploct98.htm#ant
in addition, we have never observed any other pollination candidate around any of the open pilostyles flowers.
so the great mystery of the pollinator of the pilostyles seems solved.
[Note from a later email: Bob Allen cautioned that one should not be too quick to conclude ants are the pollinators without going through the detailed 10 point checklist given in Hickman (1974), Pollination by ants: a low-energy system. I did so; we meet 6 of the 10 characteristics hands-down. we don't necessarily meet 4 characteristics, but i'm not sure they apply here, since the pilostyles flower isn't the object of the ant. it is only "accidentally" getting pollinated on the ant's journey to the p.e. flower.]
in the course of checking out the pilostyles plants, we came across two more bushes with blooms. neither of them had many blooms on many stems like the first plant in bloom; they had just ~2 flowers TOTAL apiece. the lack of bloom so far this year is quite mysterious, but it is still possible that the plants simply string out their bloom from jan to may. on the other hand, it may well be that pilostyles can only infect plants in certain years in certain locations, and so the pilostyles in the plants in this location ARE mostly dead (see below).
here is wayne's pix of one of the flowers from one of the new bushes: http://waynesword.palomar.edu/ploct98.htm#pilofe
i did a quick analysis of the data i collected that same night; the following contain some conclusions from that quick analysis.
there is no evidence that a pilostyles infection stunts the host plant. in fact, the data i collected shows that plants with pilostyles are taller than plants without pilostyles, even though the plants have the same breadth. however, i can't claim the pilostyles helps the host plants grow, since my data on host plants all come from one part of the surveyed area, and conditions might be such that those plants are simply shorter there.
my impression that many of the pilostyles infestations were dead was well-confirmed by this detailed numeric study. we found a number of plants where there was so much growth above the last dead pilostyles flower that the pilostyles was surely no longer alive in those hosts. the 3 plants with blooms all had from 10 to 35 cm of p.e. growth above the last dead previous-year pilostyles flowers. a bit over half of the surveyed plants also had that same amount of growth; these may have live pilostyles infections that may yet bloom this year or will bloom next year. (on the other hand, the pilostyles might in fact be largely dead on all plants in this area, explaining why we observe so little bloom there.)
one third or more of the plants had p.e. growth of 20-60 cm above the last pilostyles flowers. in these plants, the pilostyles are surely dead.
the vast majority of the plants had old pilostyles flowers only at the very base of the plants, on all stems, indicating that the infestation usually happens when the plant is young, and that the pilostyles flowers for one or two years, and then dies.
on one largish plant, which was therefore old, the old pilostyles flowers at its base had largely turned into almost shapeless dust, very unlike the fairly-well-preserved old flowers on other plants. wayne's pix of this plant clearly shows the pilostyles infestation is dead on this still healthy large plant:
http://waynesword.palomar.edu/ploct98.htm#olfls
on these plants, the pilostyles infestation is quite regular on all stems. some plants have 10-11 main stems, all with pilostyles. one plant has 20 stems, and clear pilostyles on 19 of those. (the 20th stem might have been infected; i didn't spend the time to study it intently after i saw 19 stems with easily-visible flowers.)
the min pilostyles length from any stem is highly correlated with the max length from any stem, meaning that the pilostyles flowers roughly the same on every stem in an infected plant.
however, we also found a few older plants that only showed pilostyles flowers on a few stems well above the base of the plant, so there are other modes for the pilostyles appearance. there are at least several possible hypotheses for this:
- the infection perhaps only began in a single stem and couldn't propagate through the plant to other stems before dying. this could be because the infection happened only after the plant was quite mature, and it was no longer possible for the pilostyles to travel to the other stems through the thick lower portion of the infected stem. (see below for a calculation of the rate of multiple infections per plant.)
- the infection was in the whole plant, but the plant was actively killing the infestation, and it struggled to produce just a few flowers before dying.
in all the plants except two, the median length of stems with visible pilostyles flowers or craters from previous flowers ranges from 5 to 25 cm. this is probably 1 to 2 years of flowering, at which point the pilostyles dies.
in the two exceptions, the pilostyles median length is 35 and 45 cm. one possible interpretation is that the pilostyles infestation lasted 3-4 years for those plants; other interpretations are also possible, including a second pilostyles infestation on those plants after the first one died.
in fact, one _expects_ the latter case, from the math (which i only realized as i was writing this note!). if ~1/4 of the plants have been infested with pilostyles, ~1/16 of the plants should have two separate infections (the exact expected numbers are only slightly different, since in this case one would observed 1/4 + 1/16 of the plants to be infected; but 1/4 + 1/16 is pretty close to 1/4, which is why i put a "~" on the previous numbers). i surveyed 32 plants; therefore one would expect two plants (= 1/16 * 32) to have two separate infections, which exactly fits those two plants observed.
one thing we did NOT find: we found no young plants showing their first pilostyles flowers. this is a bit puzzling, since there are young plants there. perhaps it takes special conditions for pilostyles to be able to reproduce from seed, and so all the pilostyles plants in an area tend to be in a few coeval generations. the special conditions might not be the _year_; it might be the location along the wash in any given year.
I thank Wayne Armstrong, James Dillane, Job Kuijt and Bob Allen for helpful comments on this article.