Plants of Southern California: Poa annua and P. infirma
Southern California Observations
Poa infirma is an interestingly little grass. If it were a human, it would have a decided inferiority problem, since in the United States, it is widely misidentified as or lumped in with P. annua. It was even originally ignored in its home region, the Mediterranean; the type specimen is from 2450 m (8000 feet) in the Andes, Columbia, South America, where it is introduced!
P. infirma was recognized in 1816, but has been ignored, or lumped with P. annua in the United States for most of the 20th century. Jepson (1909) and Hitchcock and Chase (1950) lumped them, and Munz (1959, 1974) either ignored it or didn't recognize its presence in California. Oddly, although Hitchcock and Chase lumped the two species together, their description of P. annua does not include the anther range of P. infirma, which is one of the two main distinguishing features. Munz did not include its anther range within his circumscription of P. annua either.
Robert Soreng learned about this taxon in the 1980s, and started looking for it in California. He found it in the Central Valley and central coast of California, as well as in Mexico. In his Poa treatment in the Jepson Manual, he brought P. infirma to light in California for the first time.
Having only come to botany in the post-Jepson Manual era, I diligently keyed out specimens of what looked like P. annua to see if any were P. infirma. (If I had learned botany in the Munz era, I would probably never have noticed the existence of P. infirma in California and never bothered to key out samples that looked like clear P. annua.)
The first zillion or so specimens I keyed out (well, maybe 20 or so, but it seemed like a zillion) all keyed clearly to P. annua. P. infirma thus quickly became one of my "sought-for" taxa, since I couldn't seem to find it. This was all the more surprising since Soreng's treatment said P. infirma was mostly confined to the s CA-FP.
As part of my work on the flora of the Santa Rosa Plateau, I of course had to key out every sample of P. annua I came across to make sure none were P. infirma. On 3 March 2005, I routinely keyed out a sample I collected, expecting it to be the boring old P. annua, and was dumbfounded that it keyed clearly to P. infirma.
I then checked for other records of this species in southern California, and was astounded that none of the 47 floras I had digitized at that point contained it. None of the local checklists (San Diego County, western Riverside County, Orange County) contained it; there were no vouchers online in SMASCH, and there were no records of it at UCR. This is pretty odd for a taxon listed as s GV, s CW, SW in the Jepson Manual!
I then contacted Soreng and asked how he produced his insightful treatment in the Jepson Manual. He replied that having found it in central California, and in Mexico, that it was very likely to be in southern California as well, just misdetermined.
He was right!
I have since found about half the locations sampled at the Santa Rosa Plateau have P. infirma, and about half have P. annua. In the nine locations I've sampled, I have yet to find the species intermixed. However, two of the locations are only 100 feet apart, so I would not be surprised to find them together in one location someday.
The following map shows the locations of each species at the Santa Rosa Plateau:
There are too few samples to draw any firm conclusions about any possible separation by habitat or location. With just nine locations, a human brain can almost always devise some pattern to fit the sparse points. That said, if a pattern exists, it seems that P. annua is found in the open grasslands, whereas P. infirma is found outside of the grassland areas or at their borders. (This deduction is not apparent in the above plot, since it doesn't contain the plant communities; it is from my knowledge of the locations.)
The Jepson Manual key to separate the species is:
3. Spikelet axis gen hidden, terminal internode < 1/2 terminal floret length; anthers 0.6-1 mm; widespread ... P. annua
3'. Spikelet axis visible, terminal internode > 1/2 terminal floret length; anthers 0.2-0.5 mm; esp s CA-FP ... P. infirma
The following plots show that the plants found at the Santa Rosa Plateau cleanly separate, largely as described in the above key:
I found only a single plant where, using the Jepson Manual key, the determination wasn't immediately clear upon looking at it with a microscope. On 4/26/05, along Sylvan Meadows Road between the Torino and Tovashal Trail, the first plant I looked at was indeterminate using the Jepson Manual key, with anthers 0.3-0.7 mm and a terminal internode / terminal spikelet ratio of exactly 1/2. This is the lowest blue point in the above plots. However, the anthers were nearly all 0.8 mm in length from two other samples from that same spot; in this entire set of 3 plants, I had only a few outliers shorter than 0.7 mm. The terminal internode / terminal spikelet ratio was 0.40-0.45 for the other two plants. Interestingly, the other two plants are the second and third lowermost blue points in the above plots.
It is possible that this single plant is a hybrid between the two species, but it seems more likely from the good separation in the above plots that the key simply didn't include the full range of the anther lengths for P. annua. This is expected, since measurements quoted in floras are in general based on a small number of specimens. My experience from measuring thousands of specimens from all species is that one should regularly expect to find measurements outside the numeric range quoted in the floras. Half of all the specimens I measure have at least one such measurement outside the range given for one characteristic.
Furthermore, since these species are both primarily self-pollinated, it is likely that these three plants came from parents at this location with slightly smaller than average anther lengths. See below for evidence that hybrids between these taxa should be rare, if they exist at all.
With James Dillane, I have also found this species at Daley Ranch in north San Diego County and vouchered that population. All of the above Santa Rosa Plateau populations have been vouchered as well.
P. annua and P. infirma: Species Information
Scott White perceptively pointed out that it was possible that founder effects might be responsible for plants introduced to South America differing from plants introduced to North America. P. infirma in South America might have simply come from a short-anthered parent in a European population with a range of anther lengths, whereas P. annua in North America came from a long-anthered parent. In this scenario, the defined species would in fact not be separate taxa.
Exploring this possibility led to three interesting findings:
- European floras recognize the two species as well, and they use the same anther size discriminant to separate them. The geographic ranges of the species are distinct in Europe. P. infirma has a native geographic distribution restricted to the Mediterranean and Atlantic borders, whereas P. annua is very widespread.
See, for example, floras from Northwest Europe, France, and Spain.
I also found many floras that contain both species in countries where the species are introduced.
- The two species have different chromosome numbers. P. annua is tetraploid (2n=28), whereas P. infirma is diploid (2n=14). See, for example, the Jepson Manual entries.
Furthermore, it is widely believed that P. infirma and P. supina were the ancestors of P. annua, giving it the hybrid vigor to take over first Europe and then the world. See, for example, Darmency and Gasquez 1997 and Delye and Michel 2005.
- These species are both primarily self-pollinating, and even species in this complex with the same chromosome number rarely produce fertile hybrids. (See the references immediately above.) Species with different chromosome numbers rarely produce fertile hybrids.
Furthermore, these species grow so low to the ground, normally in wet lower areas, that it makes it difficult to transport pollen from one low area to the next.
The facts combine to make it quite unlikely that hybrids will be found in any small sample such as the ones reported in this paper.
Hence it seems pretty clear that these indeed are separate species.
I will continue to collect data on samples, including measuring herbarium specimens, to see how well they separate. Introduced taxa that are able to interbreed often merge into a single taxon, if the forces that kept the taxa separate in their native land, usually geographic separation, aren't at work when they are thrown together in non-native areas.
Measurements From Herbarium Specimens
As sometimes happens, accumulating more data has thrown a little mud into what appears above to be fairly clean separation between the species. This is a progress report from my first analysis of vouchers from herbarium specimens of California plants.
On 19 January 2007, I visited the UCR Herbarium and spent 1.5 hours analyzing 28 P. annua vouchers to see if I could find any P. infirma vouchers hiding there, and to gather additional data to see how well the species separate.
I thought it would be a simple matter to look through them quickly, but it wasn't. It turns out vouchers mostly don't contain visible anthers, either because they didn't have visible anthers when collected, and/or because the anthers didn't survive the collecting and mounting process. I did not attempt to dissect any spikelets to look for anthers, since that would have been very time-consuming.
On 10 vouchers I couldn't find any anthers. On 18 vouchers, with much time spent looking under the microscope (P. annua specimens often have zillions of spikelets present, making this a needle-in-a-haystack problem), I typically found 1-2 anthers per voucher, most of which were hiding amidst the other parts. The vouchers came from a wide area, including the Central Valley, the northern Sierra Nevada, Imperial Valley, the Los Angeles Basin, the Perris Plain, Orange County and the Agua Tibia Mountains.
Note one giant caveat about all the succeeding discussion: I did not try to distinguish between fertile and sterile anthers. Most, if not all, anthers I saw were devoid of pollen, and curled. I do not know if these anthers were sterile to begin with, or curled and became empty after releasing pollen. I do know that these were essentially the only anthers available to measure on these vouchers.
On my next visit to the UCR Herbarium, I'll examine my Santa Rosa Plateau P. annua vouchers deposited there to see if the anthers have any pollen surviving after mounting. I have a dim recollection that my P. infirma samples all had sterile anthers.
The following plot shows the voucher data along with the data from the Santa Rosa Plateau:
I've now added two dark black lines to outline the area of the plot where all anthers are ≥ 0.5 mm, since 0.5 mm is supposed to be the magic separation length between these two species according to all floras.
One firm conclusion from the measurements is that it is not true that that all anthers of P. annua are ≥ 0.5 mm. I found plenty of anthers < 0.5 mm for samples with a maximum anther length of ~0.8 mm that no one could argue were P. infirma.
The separation between the two species no longer looks quite as compelling as it did with just the Santa Rosa Plateau data. As hinted at in the Santa Rosa Plateau data, the voucher measurements definitely show that there is no gap using minimum anther sizes, since the distribution is continuous across the magic 0.5 mm dividing point.
This probably isn't due to the anthers shrinking as they dry, since I remeasured one of the Santa Rosa Plateau plants after drying and found the anthers shrank by possibly ~10%, but no more. This is confirmed by fitting a sloping line to the Santa Rosa Plateau P. annua points and another one to the voucher points, and looking at the vertical shift between them. At an anther length of 1.0 mm, there is no more than a ~0.1 mm shift, also implying the shrinkage is about 10% and no more.
The following plot gives the histogram of the minimum anther sizes, combining all data and ignoring any species assignment, along with one sigma Poisson error bars on each point:
It is clearly possible to interpret the above plot as coming from two separate distributions, but there is no statistical support for doing so. One can interpret this as being a single distribution with high confidence. In fact, given those large error bars, one can fit almost anything to these data, including a nearly-flat distribution for all values.
There still may be a decent separation between the species if one uses the maximum anther length to separate them. There remains a substantial gap at a maximum anther length between 0.2 and 0.5 mm, but the Jepson Manual treatment states that the anther length for P. infirma is 0.5 mm. If the species do in fact have no gap in their maximum anther lengths, the question then becomes whether they have distinguishably different histograms of maximum anther sizes.
This is similar to Ranunculus californicus and R. occidentalis, which have a very large degree of overlap in the number of petals, but which have very different histograms of the number of petals.
The following plot gives the histogram of maximum anther size:
Note the point at (0.35 mm, 0). Its Poisson error bar is zero, so the point doesn't show up very well on this plot. (In any model fit to the data, one would use the Poisson error bars of the predicted number, not the error bar of the observed number.)
This histogram can plausibly be interpreted as containing two populations due to the apparent gap between 0.2 and 0.5 mm seen in the first plot above in this section.
However, due to the large error bars, one cannot state definitively that there is anything other than a single histogram with a tail at low maximum anther sizes. A simple single-histogram model like that fit to the data can only be rejected at the 90% confidence level, not the usual 95% level usually required for rejection. Furthermore, since the lowest points all come from the Santa Rosa Plateau population, and these are self-pollinated species for the most part, systematic error probably dominates, making the above error bars too small.
Thus I make no claim based on data I have collected so far that these species are separate and observed in southern California. Although the Santa Rosa Plateau plants with short anthers cleanly key to P. infirma, given the voucher results above, evidence for that assignment currently has to come from evidence from elsewhere of there being two species, with discrimination based on maximum anther size.
Of course, I also make no claim that the species are not separate. I only state that data gathered in California so far by myself are not definitive enough to make a claim one way or the other about the existence of these species. Presumably, there must be good data collected in Europe, or else they both would not be present in so many floras there.
In other words, if you believe the Jepson Manual treatment, which is based on floras from elsewhere, the specimens from the Santa Rosa Plateau that key to P. infirma are unquestionably that taxon. However, if you believe that floras from elsewhere have erred in separating out P. infirma, I cannot offer you any proof that you are wrong.
I'll continue to gather data on each herbarium visit to see what the California data show.
I did not in general measure the terminal internode and terminal spikelet length, since this is very time consuming and apparently is not as clear a discriminant from the Santa Rosa Plateau data. But I did make those measurements for the two specimens with minimum anther length < 0.5 mm and maximum anther length of 0.5 mm that would key to P. infirma in the voucher sample. Those measurements were:
- 1.0 mm / 2.1 mm, a ratio of 0.48, for the sample in UCR99857, with anthers 0.2-0.5 mm; and
- 0.9 mm / 2.5 mm, a ratio of 0.36, for the sample UCR174342B with anthers of 0.4-0.5 mm.
These values also key to P. infirma using the Jepson Manual key. In isolation, if one trusted that the Jepson Manual key was 100% accurate, one would conclude that these specimens were P. infirma.
However, as seen in the histogram above for maximum anther size, there is no evidence that these two vouchers should be assigned to a different species than the rest of the voucher population. I.e., they appear to be associated with the P. annua part of the histogram of the maximum anther size. Furthermore, the voucher UCR99857 had another specimen on the same sheet with an anther length of 0.9 mm, which keys clearly to P. annua. It is of course possible that the two species are found together, but one would feel better in recognizing two species here if the populations were not found in the same location.
Bottom line: I would be very reluctant to discriminate these two voucher specimens from the rest of the P. annua voucher population without a lot more measurements from samples from the same area.
The species may well be quite distinct, and the Jepson Manual key may just need some tweaking to fit California specimens. It does appear that the Santa Rosa Plateau plants are a different population from the vouchers for the anther size. More data are needed to understand how to separate these species in California.
I thank Robert Soreng for providing the interesting information about the type specimen, and the reasoning behind his perceptive and bold Jepson Manual treatment of it; Scott White for comments that stimulated the discussion here on the relationships of these species and possible hybrids between them; and Andy Sanders for his question about the significance of the separation between the species in California, and for his hospitality during my visit to the UCR Herbarium