The transition between Opuntia parryi and O. ganderi along Highway S2 in San Diego County

Acknowledgments and Final Comments


Specimens of O. parryi (=Cylindropuntia californica var. parkeri) are usually easily separated from specimens of O. ganderi (=Cylindropuntia ganderi); see Opuntia echinocarpa, O. ganderi, O. parryi, and O. wolfii: Locations and Pictorial Identification Guide which also gives additional synonyms for the scientific names. However, many specimens in San Felipe Valley and Earthquake (Shelter) Valley are not easily determined at a glance, and appear as if they were hybrids between these two species.

The possible hybridization here is on a much-smaller scale than previous speculation about intergradation between these species over a larger area. See "Intergradation" between Opuntia parryi and O. ganderi for a background on such speculation.

Specimens near Scissors Crossing on S2 in San Diego County do look like they might be hybrids between O. parryi and O. ganderi. There is a roughly 8 mile north-south strip centered on Scissors Crossing of such plants. I report here on the results of an intensive study of plants in that strip.

Prior to this intensive study, I looked at these specimens on two occasions.

The first time I was simply looking for plants that had the typical overall form of O. parryi, as seen while driving at 30-40 mph, and GPS'ing their locations. On this trip, I concluded (incorrectly, as it turns out) that there were essentially no specimens of O. parryi south of Scissors Crossing. The form of the plants south of Scissors Crossing is almost identical to that of some specimens of O. ganderi, with many erect, parallel branch tips. In contrast, plants north of Scissors Crossing have few such branch tips.

On my second visit, I made detailed measurements on specimens separated roughly every four miles. Knowing about the "suspicion of introgression" here from Jon Rebman, amidst the many clear O. parryi specimens north of Scissors Crossing, I made a special effort to sample specimens in this population that looked like O. ganderi from their dense spines on their stems. Despite their spiny appearance, analysis showed that specimens four miles north of Scissors Crossing were clearly O. parryi.

Specimens at Scissors Crossing caused considerable confusion on that visit. My companions and I were not comfortable in saying that there were two distinct species at this location. We saw specimens that seemed similar to O. parryi and other specimens that seemed similar to O. ganderi, but that was as far as we were willing to commit.

I measured in detail two specimens that day, one at each extreme in the population. Analysis of my measurements showed that one specimen was clearly O. parryi.

In my analysis I (mis-)interpreted the other specimen as O. ganderi for the following reasons. In the field, it seemed a perfect O. ganderi, with a number of erect, parallel branch tips, and was significantly spinier than the previously-seen specimens of O. parryi. In the Principal Components Analysis, it plotted away from other specimens of all species in PCA1, but close to other specimens of O. ganderi in PCA2 and PCA3. It was not unreasonable to conclude that the large separation in PCA1 was due to small number statistics, and that further data would fill in the gap with other O. parryi specimens.

However, as I accumulated more data on both species from many other locations in southern California, the gap in PCA1 filled in with O. parryi specimens. Furthermore, I was shocked to find similar very-spiny plants along the Santa Clara River in the San Gabriel Mountains, where no O. ganderi is present. It eventually became crystal clear that my detailed measurements on this specimen said the determination was actually O. parryi, even though the plant looked like O. ganderi.

This intrigued me enough that I decided to do a special survey to try to locate the precise boundary between these two species along S2, as well as to see if I could quantify any intergradation between these species which might account for the appearance of these plants.


My previous detailed measurements of specimens of these cholla species around southern California revealed that most of the separation between the species was in the tubercle length and maximum stem segment length. (This plot is given below.) In order to measure enough specimens to analyze intergradation, I decided to measure only those two characteristics for a much larger number of plants.

Measurements of just those two characteristics can reliably determine the species for about half of all specimens. (The other half have measurements consistent with either species due to partial overlap in those characteristics between the species.) Hence I decided to measure five specimens at each location, which would give me a good mean determination of the species there, as well as 2-3 reliable specific species determinations at each location.

I decided to sample 14 locations over a distance of 18 miles, with the locations determined as follows:

This design resulted in measurements on 70 plants. For each plant, I measured tubercles on at least three different stems to get a minimum and maximum value, as well as maximum stem segment lengths on at least three different stems. This meant I stooped down a total of 6 * 70 = 420 times, resulting in a sore lower back the next morning. (:-)

After the first stop, I used my car odometer to pick the next stop at approximately my desired sampling interval. The precise location for each stop was determined by spotting a good cluster of plants at a good place to stop along the road. In general, my actual sampling was within 0.1-0.2 miles of the intended sampling.

At each stop, I randomly picked plants for measurement, avoiding young plants and depauperate plants. I generally measured the closest five plants to the car. Five nearby plants could of course be genetically related, but since I was interested in the trend among these 14 locations, this was irrelevant.

The 14 locations at which data were taken, along with the complete data from each specimen, are given in a separate page.

These locations are plotted in the following maps: Large scale (area overview) and Small scale (detailed view).


In the field, I was quite surprised that I could not pinpoint the transition between O. parryi and O. ganderi. I kept expecting at the next stop I would find clear visual evidence of a transition from the morphology of the plants, but when I got to the known O. ganderi location at milepost 21.4, I still hadn't seen it. Checking my measurements, I could see that on average they had changed from previous locations, but I was fully expecting to see evidence of intergradation when I analyzed the data.

As often happens, analysis of the data surprised me. The data clearly show that there is a sharp transition between the two species, with O. parryi at milepost 20.2 and north, and with O. ganderi at milepost 21.4 and south.

This is evident in the raw data, and becomes very clear in the processed data.

The raw data are plotted in the next three plots. At each milepost value, five points are always plotted, but some values are identical and plot on top of each other. See the raw data for all individual values. In each plot, the median value for each characteristic is plotted separately for miles 10.3 to 20.2 and for miles 21.4 to 27.8.

It doesn't take sophisticated analysis to see that there is a clear break between mile 20.2 and 21.4.

For example, O. ganderi rarely, if ever, has stem segments longer than 26 cm. Six of the 10 plants at mileposts 19.1 and 20.2 have stem segments significantly longer than 26 cm, ranging up to 37 cm; only a single plant at milepost 21.4 and higher had a value longer than 26 cm, and that value was only 28 cm. Since the median maximum stem segment length for O. parryi is 25 cm, the distribution of stem segment lengths for mileposts 19.1 and 20.2 are dead consistent with all plants being O. parryi. Furthermore, it is clear just from this one plot that there can be few specimens of O. parryi at milepost 21.4 and south.

The tubercle length plots show the same abrupt transition between milepost 20.2 and 21.4.

The following plot shows the tubercle length versus max stem segment length for these plants as well as plants I have measured from throughout southern California. Those other plants have very reliable determinations made from a larger suite of measurements, and are plotted with solid (filled-in) symbols.

For the plot, I have assumed that all plants at milepost 20.2 and north are O. parryi, and that all plants at milepost 21.4 and south are O. ganderi. Analysis of this plot will then check whether that assumption is correct. I use non-solid (not filled-in) symbols to distinguish these S2 plants, with their tentative determinations here from that assumption.

The plot shows that only two of the 70 S2 specimens are inconsistent with those assumed determinations. It would not be unexpected to find a few stragglers of each species in the area dominated by the other species. If these plants are such stragglers, they have the wrong symbol associated with them in the above plot. It is also possible that these plants simply happen to have characteristics at the extremes of their species, and hence have the correct symbol in the above plot. Fortunately, the actual species assignment of these two specimens makes no difference in the analysis here.

The plot also shows that the tubercle lengths for the S2 plants are systematically shorter for both species than for plants found elsewhere in southern California. As mentioned above, this is a known effect for O. parryi, that plants found in wetter locations than here have longer tubercles. This trend is also seen in the plots of tubercle lengths versus milepost above. The plants in the north are in wetter locations, with a number of springs close to the plants at milepost 10.3.

It is also not unexpected that different populations of O. ganderi might have different tubercle lengths. The fact that the tubercle lengths of O. ganderi are smaller in this area than elsewhere in its range is also strong evidence against any intergradation with the longer tubercles of O. parryi.

Given the above, I can now probe more deeply into these data to see if there is any evidence of introgression between the species.

I use the rectangles in the above plot to tag those specimens that are clearly determined as O. parryi or O. ganderi by falling outside of the overlap of the rectangles. Specifically, any plant with tubercles shorter than 16 mm or a maximum stem segment less than 14 cm is O. ganderi. Any plant with tubercles longer than 24 mm or a maximum stem segment longer than 26 cm is O. parryi. The following plot shows the number of clearly-determined species from each location.

This plot shows no evidence of any intergradation at all. If there was intergradation over this area, the number of clear determinations of each species should be small at the zone of maximum overlap between the species, and rise to each end of the plot. This is completely inconsistent with the data. Instead, the data show a sharp transition between the species between mileposts 20.1 and 21.4.

The following plots investigate whether there is more subtle evidence of intergradation by taking the median of the data at each milepost.

The median maximum stem segment length versus milepost in shown in the following plot:

Also shown are:

Once again, there is strong evidence that there is no intergradation between the species here. There is no gradual transition in the median maximum stem segment lengths with milepost number expected if there was intergradation. Within measurement error, the black lines are consistent with the red lines, implying there is nothing different about these plants from the same species in other areas of southern California.

Finally, the following plot shows the median tubercle lengths:

Although the median tubercle lengths for O. parryi appear to show intergradation with O. ganderi, the interpretation of this plot is complicated by the known variation in tubercle length of O. parryi with moisture. The gradient from north to south in the O. parryi median tubercle lengths is consistent with that known variation, and so one cannot use this plot as evidence for or against intergradation.

Fortunately, there is no such effect in the tubercle lengths of O. ganderi, so its portion of the plot can be validly examined for intergradation. Once again, no intergradation is evident. Any possible slope to the O. ganderi points is in the opposite sense.

Simulation To Determine Sensitivity to Intergradation

In order to determine how sensitive my data were to any possible intergradation, I performed the following simple experiment. I randomly switched two sources between mileposts 20.1 and 21.4, and a single source between mileposts 19.1 and 22.4. This had the effect of artificially introducing intergradation in the dataset, which I can then analyze as above in order to see if I could detect intergradation between mileposts 19.1 and 22.4.

The following plots are identical to the above, but now using the simulation data:

The plots now show intergradation. The number of clearly-determined species on the "wrong" side of the boundary has doubled from 2 to 4, and the plot clearly shows the effect at the boundary. The median stem segment lengths at milepost 20.2 and 21.4 have now become nearly identical at intermediate values between the species. The median tubercle lengths near the boundary have now also become nearly identical, showing clear intergradation.

I conclude that my data are easily sensitive enough to pick up any intergradation in this area. The fact that no evidence was found is therefore strong evidence that no such intergradation exists, at least in the measured characteristics.


Acknowledgments and Final Comments

My work here would not have been possible with the considerable body of work of Lyman Benson, Bruce D. Parfitt, Marc A. Baker, Donald J. Pinkava, and Jon Rebman. The genus Opuntia has been tremendously confusing to sort out taxonomically, and all of these people have made considerable contributions. My contribution here has been to intensively study just a small part of one of the remaining puzzles.

These cholla species are often sufficiently similar that it is very easy to get fooled by a voucher specimen, and even easy to get fooled when one sees live plants in a new area for the first time. Nearly every time I study a new area, I have gotten fooled by at least one specimen when I haven't spent the time to study it thoroughly and think carefully about the determination in the field. Thank goodness numbers are not so easily fooled! (;-)

I especially thank the botanists and friends who have studied these plants on long car trips with me, including James Dillane, Paula Knoll, Norrie Robbins, Kate Shapiro, Dave Stith, Jane Strong, and Dick Swinney.

I thank Jane Strong for excellent comments on this manuscript.

Go to:

Copyright © 2007 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: 28 February 2007