Plant Species of San Jacinto Mountain: Arabis = Boechera species with non-erect fruit: Background Information

This page is an older page that gives background information on these species, prior to analysis work done on these species at San Jacinto. As a result, some of the comments below have been superceded by later work given on the latest page on these species.

Voucher Information
Species Characteristics
Recent Changes in Delineation of These Species
Recent Windham - Al-Shehbaz Voucher Determinations
Conclusions on Species Determinations


This page discusses only the Arabis species that have non-erect fruit and that are found in the San Jacinto Mountain area. The Arabis species with erect fruit are much more easily separated, and will be ignored here. All species not found in the San Jacinto Mountains area are also ignored here.

Boechera is the current genus name for these plants. DNA work has shown these species diverged some 19-25 million years ago from Arabis species in the Old World. (After all, the name Arabis came from Arabia!) I will use both names in this article.

First, a giant disclaimer: I am not an Arabis expert. I have pathetically-little experience in studying the Arabis species in southern California in any detail other than trying to identify the ones I have come across.

In fact, for several years, I have been greatly confused by the Arabis species with non-erect fruit in the higher-elevation (above ~6000 feet) San Jacinto Mountains, for two main reasons:

I am still collecting data on these plants, and I have yet to do any detailed analysis on more than a handful of specimens to try to understand what is going on here. However, as a result of careful study of the recent papers by Windham and Al-Shehbaz, pollen size measurements made by Wayne Armstrong on these plants, and the voucher analysis reported in this page, I now understand a lot more about the issues here.

It appears that the reason for my confusion is that these plants in fact are not any of the species described in Munz and the Jepson Manual. Instead, they are one of the many apomictic species of hybrid origin in this genus, which apparently outnumber the sexual diploid Arabis species. (Apomixis means asexual reproduction, without fertilization. This also nicely explains my puzzlement over seeing what appear to be fresh unfertilized flowers, but with fruit already developing.)

I am apparently not the only one confused about these specimens, since Windham and Al-Shehbaz state that misidentification rates in herbaria run as high as 40% for Arabis vouchers. See below for more information.

The good news here is that it looks like I finally understand this San Jacinto species, and I am deeply grateful for all the work Windham and Al-Shehbaz have done on this genus.

However, the bad news here is that Windham and Al-Shehbaz not only have no binomial name for this species, and most similar species, they recommend that such species do not receive a binomial name. Their policy derives from a good theoretical motivation to keep the sexual species separated from the apomictic species. However, this policy means that those of us who create floras have to list these true species by an unwieldy name like Boechera perennans X unknown other parent, or worse, Unknown Boechera parent X unknown other Boechera parent since most of us will not even have a clue as to whom the parents are. Without much detailed work, including chromosome and/or DNA studies, the name is just a guess, and it may change with time as more is known, or as other people make their guesses.

Furthermore, those who use floras who are not familiar with this unusual custom will perceive that this species is somehow different from other species in a number of ways, even though this species behaves just like most other species in an area.

All of this seems like a giant step backward from the beautiful simplicity wrought by Linnaeus.

This policy doesn't affect just one species found in an area of ~100 square miles at San Jacinto Mountain; it affects widespread, well-known taxa such as Arabis sparsiflora var. californica = Boechera californica, which is also an apomictic species. Had this species not already been named, this policy would require that this species be called Boechera arcuata X B. pulchra, even though this is a widespread species in southern California, and does not consist of F1 crosses of those parents. (One problem here is that the same X notation is used by botanists for an F1 cross as for an ancient species that originated by a cross millions of years ago which has since experienced much evolutionary change.)

The original purpose of this page was just to collect in one place some of what is known about these species here, which many readers may find as surprising as I have. As often happens, enlightenment sometimes results from that simple process.

Since the analysis here begins with voucher information, and most vouchers retain the species names given in Munz and the Jepson Manual, those names will be used in the next two sections.

Voucher Information

Five species (six taxa including varieties) have been vouchered from the San Jacinto and Santa Rosa Mountains. The following table summarizes the online voucher information obtained from a search of the Consortium of California Herbaria on 4 November 2008. All voucher locations used as the basis of this summary are shown in the plots below.

Species# VouchersElevation Range (feet)Geographic AreaLocations
Arabis holboellii65060-8200Higher-elevation west side, Toro PeakIdyllwild to Tahquitz Ridge, Toro Peak
Arabis perennans13600-7900Higher-elevation west side, all elevations desert sideIdyllwild, Tahquitz Valley, Santa Rosa Mountain, Palm Springs
Arabis pulchra81000-6300All areas except higher elevationsJames Reserve, Pine Cove, Garner Valley, Thomas Mountain, Palm Canyon, Whitewater
Arabis repanda26440-6700Tahquitz AreaTahquitz South Ridge Trailhead, Tahquitz Valley
Arabis sparsiflora (varieties arcuata and californica)382400-7000Mostly south and west mid-elevations, Santa Rosa MountainsBautista Canyon, Poppet Flat, Fuller Creek, Pine Cove, Idyllwild, Garner Valley

Although varieties are given for Arabis pulchra and A. holboellii in both Munz and the Jepson Manual, I will ignore those varieties for two reasons. First, all vouchers from here of A. pulchra are just determined to the species. Second, three of the six vouchers of "A. holboellii" are determined as var. pendulocarpa, which is not present in southern California according to both Munz 1974 and the Jepson Manual. Further, as will become apparent below, A. holboellii (for all specimens) and A. perennans (for some specimens) are probably not the correct names for the specimens currently so determined.

Of the 38 Arabis sparsiflora vouchers, 13 are determined just to the species, five are determined to var. arcuata, and 19 are determined to var. californica.

Voucher locations are given in the following three plots. The first two plots are maps, the first with landmarks (roads, streams, elevations, features) for orientation, and the second without landmarks to show the voucher locations more clearly. The third plot shows elevation versus longitude to show the elevation range of vouchers versus coastal / desert side of this area.

Of the 67 vouchers retrieved from the Consortium on 4 November 2008, 27 had coordinates, 22 of them from UCR. I georeferenced 35 vouchers that had localities and / or elevations detailed enough to plot them with reasonable accuracy. Five vouchers are not plotted in the following maps since their locations were too imprecise, such as Snow Canyon with no elevation information.

In interpreting the following plots, keep in mind that the accuracy of voucher locations, including those that already had coordinates, varies considerably. Of the 62 vouchers plotted, 37 are located fairly precisely. The remaining 23 plotted vouchers had enough information to plot their general locality, such as nw slope Santa Rosa Mts., elevation 7000 feet, but are not precisely located.

Map of Voucher Locations With Landmarks

Map of Voucher Locations Without Landmarks

Voucher Elevation vs. Longitude

The plots above show that two of the species have coherent geographic ranges:

The other three species either show less coherent ranges, or have too few vouchers to draw conclusions about their range.

Species Characteristics

See Appendix 1 for a modified Munz / Jepson Manual key to these species.

In order to see how distinctive these six taxa are, I compiled the main features of the stems, leaves, hairs, pedicels, sepals, petals, and fruit in a spreadsheet. The spreadsheet was not complete for every species for every characteristic; for example, neither Munz nor the Jepson Manual gave the sepal length for A. perennans. I then highlighted characteristics that were found in only one or two taxa, and tallied up how many highlighted characteristics each taxon had.

A. repanda had ten such highlighted characteristics; A. pulchra had nine; and the other four taxa had only a single highlighted characteristic each. Thus A. repanda and A. pulchra are very distinctive species, and it seems there should be no trouble distinguishing them from the others (but see below...).

Most of the six taxa have sessile narrow non-linear leaves, with the cauline leaves auricled; pink to purple petals shorter than 14 mm; fruit less than 80 mm long and 2 mm wide, spreading to reflexed, uniseriate seeds, and seeds less than 1.5 mm wide, with wings less than 1 mm wide.

A. repanda and A. pulchra each differ from the other of the six taxa as follows:

A. repanda: basal leaves petioled, repand-toothed, 3-8 cm long, 1-3 cm wide; lower cauline leaves with winged petiole, upper cauline leaves sessile, not auriculate; petals white to pinkish; fruit 40-100 x 2-4 mm; seeds 2.5-5 mm long, 2-4 mm wide (including wing); seed wing (0.5?)-3? mm wide).

A. pulchra: basal and cauline leaves linear, sometimes hairy enough to appear covered with a white down; cauline leaves sometimes not auriculate; petals 8-20 mm long; mature fruit spreading, pendent, or reflexed-appressed, 40-70 mm x 2.5-3.5 mm; seeds biseriate.

The remaining four taxa have the following characteristics in which they show some differences:

TaxonBasal leaf size (mm)Petal colorPedicel Length (mm)Pedicel ShapeFruit size (mm)Seeds
A. holboellii10-503-6whitish to pinkish6-12spreading and arched downward or recurved to +- reflexed35-801-2mostly uniseriate
A. perennans20-604-20pinkish to purple10-20spreading and arched downward40-601.2-2uniseriate
A. sparsiflora var. arcuata30-1002-12pink or purple5-15spreading to ascending60-1201.5-2uniseriate
A. sparsiflora var. californica30-1003-13pink or purple5-15spreading-recurved60-1201.5-2uniseriate

In the above table, the pedicel characteristics refer to those for mature fruit.

Note that the key given previously is not consistent with the entire range of pedicel characteristics of A. holboellii! The straight pedicel refers only to var. retrofracta in Munz; var. pinetorum has pedicels arched downward, exactly the same as in two of the other three taxa. It seems very difficult to separate A. holboellii var. pinetorum from the other three taxa.

In fact, as discussed below, it appears that A. holboellii, and possibly A. perennans, do not actually exist in this area, which makes most of the above discussion of the properties of those species mute for these plants here.

Recent Changes in Delineation of These Species

Ground-breaking work by Windham and Al-Shehbaz, in two papers in Harvard Papers in Botany (11: 61-88, 2006; 11:257-274, 2007) have made it clear that the taxonomy of these species is, essentially, a taxonomist's worst nightmare. They state:

Arabis exhibits a rare confluence of hybridization, apomixis, and polyploidy that makes it one of the most difficult genera in the North American flora. The sexual diploid species are relatively distinct from one another, but they hybridize whenever they come into contact. Through apomixis and polyploidy, these hybrids become stable, self-propagating lineages. Most of the hybrid derivatives in Boechera are triploids, but apomictic diploids are known as well. Thus, for any pair of sexual diploid species (e.g., AA and BB), this process can yield several different intermediates, including AB apomicts and both possible apomictic triploids (AAB and ABB). The situation becomes even more challenging when a third sexual diploid enters the picture. To date, we have identified several taxa that appear to be trigenomic triploids. Under these circumstances, even the most distinctive sexual diploid progenitors can become lost in a seemingly continuous cline of morphological variability.

Apomixis means asexual reproduction, without fertilization.

Part of the nightmare for plants in southern California is highlighted by the following.

First, the taxon A. holboellii, including all of its varieties, apparently does not exist in southern California.

In fact, no modern flora gives A. holboellii, or any of its varieties, in southern California. The Jepson Manual gives it only from the Sierra Nevada north. We in southern California often assume that the Jepson Manual author simply made a mistake when southern California was omitted from the distribution of a well-known species here, but that is not always the case!

Windham and Al-Shehbaz have found that two of the former varieties of A. holboellii are actually separate species, Boechera pendulocarpa and B. retrofracta. These species have no close relation at all to the true A. holboellii, which they state is currently known only from Greenland! These two species apparently are still only found from the Sierra Nevada north, although Windham and Al-Shehbaz have determined one specimen from Toro Peak in the Santa Rosa Mountains as an apomictic hybrid species with one parent being B. retrofracta. (Remember these are ancient hybrids, and hence this does not imply that B. retrofracta is still in the neighborhood.)

The third variety of A. holboellii given for California in the Jepson Manual is var. pinetorum. Windham and Al-Shehbaz state that this name was originally applied only to plants confined to the northern Sierra Nevada and adjacent southern Cascades, and they have restricted the name Boechera pinetorum to refer only to those plants. They state that the name of A. holboellii var. pinetorum morphed into a polyphyletic "supertaxon" extending through central and western North America, and was a catch-all for a large number of unrelated taxa.

The root problem, according to Windham and Al-Shehbaz, is that taxa given this name apparently are of hybrid origin with one genome from a widespread taxon, Arabis holboellii var. retrofracta = B. retrofracta, along with genomes from different parents in different areas, which became stabilized apomictic species. These apomictic species tend to look similar due to the B. retrofracta gene contribution, and hence were conveniently lumped together under the name A. holboellii var. pinetorum. Often those stabilized species then replaced the parent species in the area where those parent species formerly overlapped.

Second, Windham and Al-Shehbaz state:

The current classification of Boechera does not meet the needs of either the scientific community or the interested public. With misidentification rates in herbaria running as high as 40%, the names currently applied in Boechera are losing any predictive value.

The following table shows examples of the misdetermination rate for these taxa. For this table, I compiled the determination history for all of the taxa considered in this paper, which is given for the UC/JEPS vouchers online. The table gives the determination history for those vouchers with changed determinations.

The following table gives the vouchers of these species from southern California, from UC / JEPS, with changed determinations. The original determination is numbered 1, with subsequent determinations numbered sequentially.

Vouchers with changed determinations

Arabis holboellii   1       1  21
Arabis perennans 2  1   21  1 12
Arabis pulchra     1       1  
Arabis pulchra var. gracilis          1     
Arabis sparsiflora        3       
Arabis sparsiflora var. arcuata11, 3  2, 3 111       
Arabis sparsiflora var. californica24  32224       
Boechera retrofracta x unknown other parent   2       2    
B. perennans x unknown other parent         22     
Boechera xylopoda = A. pulchra var. gracilis            22  

All determinations made in the last four rows were made by Windham and Al-Shehbaz in 2006 or 2007, which is why A. pulchra var. gracilis appears in two rows. The vouchers are identified by a letter in this table to make the columns narrower; Appendix 2 gives the correspondence between letter and voucher name. Voucher E turned out to have two separate taxa on a single sheet, which is why it ended up with two different determinations. Windham and Al-Shehbaz made both determinations for that voucher.

Two vouchers have additional information in the determination field. Voucher F has annot. written as "prob. > A. sparsiflora Calif." in pencil just above collection label. Voucher J has a question mark after the first determination.

A summary of the changes is as follows:

It is not surprising that taxa originally determined as A. holboellii or A. perennans have become other determinations by Windham and Al-Shehbaz, since A. holboellii and non-desert-area specimens of A. perennans are not found here according to them, and thus those determinations were therefore suspect.

But two sets of these changes are very surprising:

Finding such changes means that it is very risky to rely on current voucher determinations made by anyone other than Windham and Al-Shehbaz in 2006 or later, which are detailed in the next section.

Recent Windham - Al-Shehbaz Voucher Determinations

There are 33 online vouchers from southern California that have been determined by Windham and Al-Shehbaz in 2006 and 2007. The following table summarizes those determinations:

Taxon# Vouchers
Boechera arcuata1
Boechera californica14
Boechera perennans4
Boechera perennans X unknown other parent1
Boechera pulchra / B. xylopoda4
Boechera pulchra X unknown other parent2
Boechera repanda4
Boechera retrofracta X unknown other parent2

In addition, there is one voucher curiously determined as B. sparsiflora which will be ignored in the following.

Nine of these vouchers had coordinates, which I reviewed for accuracy. I corrected the location of one of those and I added missing elevations to another six of them. I georeferenced all of the remaining vouchers except one voucher of B. repanda, which had a location only as San Bernardino Mountains with no elevation.

The following three plots show the locations of those vouchers by determination, first for all of southern California and then zooming in twice to get to the San Jacinto Mountain area. The fourth plot shows elevation versus longitude for all vouchers. (Note that there are two sets of vouchers with the same locations, so there are fewer than 32 separate points in these plots.)

Map of Voucher Locations Determined By Windham and Al-Shehbaz in 2006 and 2007 - All of southern California

Map of Voucher Locations Determined By Windham and Al-Shehbaz in 2006 and 2007 - San Bernardino Mountains, San Jacinto Mountains, Peninsular Range

Map of Voucher Locations Determined By Windham and Al-Shehbaz in 2006 and 2007 - San Jacinto Mountain Area

Elevation vs. Longitude for Voucher Locations Determined By Windham and Al-Shehbaz in 2006 and 2007 - All of southern California

Just as seen in the plot of all vouchers given previously, Boechera californica is the most common taxon, with a coherent geographic range on the coastal side of the mountains. B. perennans appears now to have a coherent distribution at elevations below 5000 feet in the desert.

The plots give a hint that the common high elevation species in the San Jacinto / Santa Rosa Mountains and San Bernardino Mountain might be B. retrofracta X unknown other parent. The only other voucher above 7100 feet is B. repanda, which is a very distinctive species clearly separate from most of the taxa considered here, and very rare in the San Jacinto / Santa Rosa Mountains. (I have yet to find a specimen of B. repanda there, despite extensive field work.)

One of those two vouchers of Boechera retrofracta X unknown other parent was found at Toro Peak in the Santa Rosa Mountains. The pollen size measurements made by Wayne Armstrong on a specimen in Round Valley at San Jacinto Mountain reveal that at least this specimen in Round Valley is an apomictic hybrid species. My observations suggest that all the specimens at San Jacinto Mountains above ~6500 feet are identical to this Round Valley specimen. Although I have not yet studied the plants at Toro Peak to see if they are identical to the San Jacinto Mountain plants, it is quite possible that they are the same.

However, until the San Jacinto Mountain plants are compared to the Toro Peak plants, it is also possible that the San Jacinto Mountain plants are an apomictic hybrid species involving a different set of parent species.

Conclusions on Species Determinations

My conclusions on species determination of the Arabis / Boechera species that have non-erect fruit, found in the San Jacinto Mountain area, are:

Voucher data provided by the participants of the Consortium of California Herbaria ( I especially thank UC/JEPS for providing the determination history of their vouchers online. I thank Steve Boyd for telling me about the Windham and Al-Shehbaz papers, which led to the analysis reported here.

Go to:

Copyright © 2008 by Tom Chester.
Commercial rights reserved. Permission is granted to reproduce any or all of this page for individual or non-profit institutional internal use as long as credit is given to me at this source:
Comments and feedback: Tom Chester
Updated 4 November 2008 (reference to latest results added 30 May 2011).