Plants of Southern California: Some Things You May Not Have Known About Vouchers

Introduction

A voucher (pronounced vow' chur) is a plant collected from a known location, with enough parts present to identify it, that is pressed, dried, and then stored in an herbarium (see Preparation of Plant Specimens for Deposit as Herbarium Vouchers). This sounds like a terrible thing to do with a plant, but pressed dried plants are sometimes still a thing of beauty, and retain much information about the plant. Some vouchers are approaching 500 years old, and 100 year old vouchers can still look similar to vouchers collected recently.

Vouchers are extremely important to botany:

Vouchers are the foundation of botany:

I'm a big fan of vouchers, as well as the people who collect them, and consider herbaria to be holy places with a treasure trove of information. It is a real joy to examine different species side by side using vouchers, so one can see with one's own eyes how the species differ, as well as the range of variation of an individual species. I wish I lived closer to a herbarium; there are so many things I could learn! There is more than a lifetime of analyses one can do on vouchers to answer different questions about plants.

However, as wonderful as vouchers are, they have their limitations, which are clearly not well known, since many people have a number of misconceptions about vouchers. This page resulted after two people wrote me in a single week that they were surprised that the voucher flora of an area they were interested in was so incomplete.

Some Things You May Not Have Known About Vouchers

Each of the following topics is discussed in a separate section below:

Voucher Records Are Haphazardly Distributed Across California

Some people think that botanists have long ago scoured every area of the state, sampling the flora in every nook and cranny. Nothing could be farther from the truth.

Dick Moe has summarized this succinctly: Most taxa and most places are undercollected.

In a small number of areas, botanists working on the flora of a given area, have indeed tried to voucher all the species in a given area. Such areas make up a miniscule fraction of the entire state, probably something on the order of 1% or less. Thus the odds of anyone finding such a complete record for a given area of their interest is less than 1%.

A quick way to tell whether you are lucky enough to be interested in an area that has a fairly complete voucher set is to see if there is a published flora for that area. If there is no published flora for that area, there is essentially no chance that the voucher records are highly complete.

Most vouchers result from one of the following sources:

A map of all georeferenced vouchers in the Consortium is found on their About the Consortium page (see also the link from the voucher map giving more detailed information). Note that the distribution of vouchers is very far from uniform, with most vouchers collected near where most people live, which is not terribly surprising. Many areas of the state have no or few vouchers, mostly in remote areas. Some areas, such as National Parks, have very few vouchers, since for many years botanists were discouraged from collecting in those areas and not given permits to collect. Joshua Tree National Park was a black hole in voucher distributions even as of a few years ago; large parts of it still have no vouchers at the Consortium.

A few botanists make attempts to identify similar black holes in areas near them, to see if they can get any exciting finds (like new species!) in them. But there are more black holes than there are botanists.

The Completeness of the Flora of a Given Area From Vouchers Depends Strongly On Area

The Consortium of California Herbaria has over 2 million specimen records from over 30 institutions as of 28 February 2014. People think that surely that means any given area has a complete record of the flora of that area.

In fact, the completeness of the flora of a given area from vouchers is a strong function of the size of the area. Very small areas typically have zero vouchers; the completeness is zero. The reason is simple: the area of California is large, and the number of botanists collecting vouchers is small. Most areas have never had a visit by a botanist collecting vouchers.

Fig. 1 shows an example of an area someone asked me about on 23 February 2015. That person was extremely surprised that there were zero vouchers in the area she was interested in! That is just one of zillions of black holes in voucher locations.

Fig. 1. A map of vouchers from a geographic search in the Murrieta, Riverside County region for the area of interest outlined in red. The numbers in the circles given the number of vouchers found in a small distance from the center of the circle. Note that not a single voucher was taken in that area of interest.

The Consortium gives a graph of the number of vouchers per square mile by county. Most counties have 10 to 20 vouchers per square mile, with the lowest being Kings County at 0.9; Merced at 3.4; and San Joaquin County at 3.5 vouchers per square mile; and the highest being San Francisco at 125, Marin at 43, San Mateo at 41, and San Diego at 41 vouchers per square mile. The number of vouchers per square mile is clearly a strong function of the population density.

This means that an average square mile will have just 10 to 20 vouchers, yet an average square mile will contain 100 to perhaps 1000 species! Many areas of interest are just a few square miles, and will contain fewer than 100 vouchers. The total number of species from those vouchers will be fewer yet, since there almost always is duplication of species in such sets of vouchers, and often botanists collect multiple voucher sheets of a single species, to share or trade with other herbaria.

As the area increases, the completeness of the voucher flora goes up as well. For areas of 50 square miles, one expects on average 500 to 1000 vouchers. The species list from that large number of vouchers is typically 80 to 90% complete.

The voucher flora for very large areas is usually close to 100%. For example, even though many small areas of the San Gabriel Mountains have very incomplete floras, there are a large number of vouchers from the San Gabriel Mountains as a whole, almost guaranteeing that few species that occur there have gone unvouchered.

As an example, if you are interested in one canyon in the San Gabriels like Fish Canyon, its voucher flora might only be 50% complete. But the voucher flora of the entire San Gabriel River Canyon and its tributaries, of which Fish Canyon is only one small tributary, might be 90% complete, and the voucher flora of the entire San Gabriel Mountains region might be 98% complete.

Actual numbers from some floras I've done for the Borrego Desert are:

The percent of taxa from those areas that are vouchered are 0, 7, 54, 87 and 80%, respectively. These numbers imply that Borrego Desert areas small enough to have only 100 species or so are very unlikely to have any vouchers at all, but in larger areas that contain 150 to 300 species the voucher flora is 50 to 90% complete.

Due to the strong dependence of the completeness of vouchers with area, when making a flora of an area from vouchers one has to make a trade-off between completeness and reliability. That is, one can expand one's search to include vouchers from neighboring areas. That will increase the completeness of the flora compiled from vouchers for the area of interest, but at the cost of putting species in that flora that aren't actually found in the area of interest. If one is more interested in completeness, to see what species might be found in a given area, this is the right choice.

Vouchers Are Primarily From Roadsides

Most people never venture far from their car while on a trip (the average length of time for people visiting the Grand Canyon is just five to seven hours!), and botanists are no exceptions. It is also far easier to collect along roadsides, especially if one wants to sample several locations in a day and select a good area to sample by sighting it from the car, and because a plant press is not a fun thing to carry around.

The vast majority of vouchers are taken from roadsides and trails. See, for example, Voucher Coverage for the San Jacinto Mountain area. Fig. 1 also shows an area where vouchers were taken entirely from roadsides.

Common Species, Hard-to-Voucher Species, and Typical-looking Species are Rarely Vouchered

Not surprisingly, common species are rarely vouchered. There is little incentive to do so unless one is doing a flora of an area. Herbaria are filled with numerous specimens of the most common species, and it mostly just takes up space to add another voucher of a common species. Furthermore, botanists know what the common species are, so they are never vouchered to get an identification.

Erik Blume was very frustrated when he first tried to put together a flora of Garner Valley from vouchers in 2006 since chamise, Adenostoma fasciculatum, was not vouchered from there at the time, even though it is a common plant there! (In a search on 2/28/15, there are now two vouchers of it there.)

One exception is when a common species "looks different". Tommy Stoughton tells the story of him coming across some very tall plants of Bromus tectorum, cheat grass, much taller than it was normally. He collected some specimens to compare it to vouchers to see if it might be some other species. Much to his surprise, when he looked at the vouchers, he found that a large number of them were of very tall plants just like his. I.e., a number of other botanists had done exactly what Tommy had done. Note that this sampling of the unusual creates a bias in herbarium specimens toward unusual specimens, which is not necessarily a bad thing as long as one is aware of that bias.

Species that are hard to voucher are infrequently vouchered even by professional botanists. Cacti are the most prominent examples of plants that require special effort to make a pressed dead plant. In addition to the general nastiness of handling and trying to press a specimen with prominent spines, unless one follows a proper procedure, cacti can continue to grow, or will mold and decay while pressing. Poison oak is also a species that few people want to voucher.

Plants with bulky or hard-to-reach parts, like pine cones, mature palms, are also vouchered less frequently.

And, of course, plants that don't grow along roadsides are less-frequently vouchered.

Dick Moe's List of Biases in Vouchers

Dick Moe kindly provided a excellent list of biases in vouchers that he has noted:

Dick also points out that biases can arise from each of the following steps in the process of creating an online voucher:

Dick notes that each step has its own set of biases or its own filters, and that each collector probably has a unique bias profile.

Voucher Coordinates and Elevations are Rarely Very Accurate

People who georeference vouchers are high on my list of people doing good things. Now that most vouchers are georeferenced, I hardly ever look for vouchers that are not georeferenced; they are becoming invisible (just like vouchers that haven't been digitized and placed in the Consortium are essentially invisible now).

However, it is nearly impossible to digitize most vouchers with a great deal of accuracy. If a voucher location is just San Diego County, where are you going to georeference it? The same applies to a voucher locality of San Jacinto Mountains; that's a big place, and it makes a big difference whether you have a high-elevation species found only above 8000 feet or a desert species on the low-elevation desert side of that area.

Most older vouchers give just vague locations like the preceding. After all, if you are from UC Berkeley in the year 1920 out collecting in the desert, all you may care about is that a voucher was collected in the "Sonoran Desert", and that may be all you put on the voucher label (or maybe "west edge of Sonoran Desert"). But if you are doing a flora of the Borrego Springs area of the Sonoran Desert, you are frustrated by such vouchers, whether they are georeferenced or not, since they may or may not be in the Borrego Springs area.

Furthermore, many locations don't have a place name. If you are out in the Sonoran Desert, a collector might just give the locality of the nearest town. Hence there are many vouchers that say Borrego Springs, or Idyllwild, that were collected 30 miles away from either location. These have to be georeferenced at the stated location, but many of those species won't be found there.

Worse, there is often historical confusion about place names, with the place names in vouchers not the same as place names known today. See Borrego Desert: Historical Confusion about Blair Valley, Pinyon Mountain and Whale Peak, which also mentions that Hellhole Canyon was applied to modern-day Tubb Canyon in a 1942 topograph map.

The other complication here is that many people on vacation don't really know where they are very precisely. So they might state a definite locality, but that's not where they really were. Or they might be confused about the name of the locality. One example I just came across, while searching for chamise vouchers in Garner Valley as mentioned above, is a voucher of chamise from San Jacinto Peak. There is no chamise anywhere close to San Jacinto Peak; as far as I know from my surveys and from vouchers, no one has ever found chamise above 6000 feet there, let alone at 10,800 foot San Jacinto Peak. The person must just be using the term San Jacinto Peak to mean someplace in the San Jacinto Mountains.

Elevations given in the voucher itself, or in georeferencing, are even more uncertain. If you don't know where you are exactly, you can't get a good elevation. Many collectors don't have a topo map of the area where they are collecting, and are just guessing at the elevations. For example, on his trip to the Borrego Desert Area in 1928, Willis Linn Jepson, one of the premier botanists ever in California, wrote in his Field Book that his elevations were all a Rough estimate. All locations in desert similar estimates, where no contour maps.

Andy Sanders did a study once of the accuracy of elevations in vouchers, and found that they were the second most unreliable voucher label item (the first being the determination!). I've done a similar study of a smaller number of vouchers, and have found many vouchers, including ones by Munz at San Jacinto Mountain, are off in elevation by 1000 feet or more. There are also a small number of vouchers where meters and feet have gotten confused, resulting in a factor of 3.28 error in the elevations.

Remember, GPS units were not widely used by botanists until the last decade or so. GPS locations are not immune to error, since if a unit is turned off and on only at each site, the user might not have waited until the GPS unit found the new location, and errors in recording a string of five or six digits after the decimal point are not infrequent.

Georeferenced positions also can contain typos. I typically throw out a few vouchers in most sets I use to make the voucher flora of an area which have big typos in their coordinates. About one in three times that I make plots of the geographic distribution of a given species for San Diego County some vouchers will plot hundreds of miles outside of San Diego County due to such typos.

Vouchers Records Are Only As Good As The Collector

For years, I'd wanted to botanize Mountain Spring Grade at the border of San Diego and Imperial county due to all the interesting species collected there that I'd never seen in San Diego County. But while getting ready for my first trip there, and actually looking in detail at the vouchers from there, I realized that nearly all of these curious plant species are from a single collector, Charles Russell Orcutt, from what appears to be a single collection event in April 1917.

A bit of investigation then found that Orcutt had collected vouchers from a very large area, probably from the Mojave Desert to Mountain Springs Grade to the coast of San Diego County, and called the entire collection his Mountain Springs Collection. So all of his curious vouchers, which of course are georeferenced as being from Mountain Springs, are actually from a hundred or more miles away! By the way, Orcutt was said to be a librarian's nightmare, somewhat careless in his work, and careless in his collecting methods. He was almost surely not the only one to which those labels could be applied. See Flora of Mountain Springs Grade area for more information.

Marcus Jones is another botanist whose locations are sometimes suspect, collecting so many specimens at once that he mixed up some of his locations in assigning them to his vouchers.

It is not uncommon for botanists to put a number of specimens in a bag as they travel, and try to sort them out into localities in the evening, or even a few days later. This is undoubtedly a source of locality error in vouchers even now.

The Consortium maintains a very nice set of curatorial tools where one can find vouchers that have the same collector and date, with varying location that is 60 to over 240 miles apart. Some of those collectors may well have collected in widely separated locations, but some of these are surely errors in the voucher labels.

Voucher Determinations Can Be Incorrect

There are a number of ways that a voucher determination can be wrong:

There are other reasons for voucher misdeterminations as well, which include typos in digitizing voucher labels, and putting the wrong voucher label on a given specimen (I've seen a switch of two labels that each went to the wrong specimen).

The Consortium maintains a wonderful list of duplicate vouchers that have different names (see Heteroduplicates in the curatorial tools). On 1 March 2015, there were 38,186 sets of such duplicate vouchers! This is 7% of the total of ~271,000 duplicate vouchers.

It is not possible to infer from that 7% number what the actual percentage of species with taxonomic problems is, for at least four reasons. First, as detailed below, heteroduplicates might arise from other reasons than taxonomic difficulties. Second, the number of vouchers is not the number of species, so one would have to go through the heteroduplicates and compile the species list from them. Third, problematic species might be vouchered more frequently. Fourth, duplicate vouchers often retain the original determinations, or might all change to a new determination simultaneously. They can only become heteroduplicates when someone redetermines the voucher at one institution, without feeding back that determination to the other institutions with duplicates of that voucher.

That Consortium page gives the following reasons for these heteroduplicates:

Why might things get labeled as duplicates that aren't duplicates? Can we do something? Why might things that are apparently duplicates be stored under different names? Can we do something?

Voucher Data provided by the participants of the Consortium of California Herbaria (ucjeps.berkeley.edu/consortium/).

I thank Dick Moe for his excellent additions to this page, Andy Sanders and Tommy Stoughton for permission to use their quotes here, Andy Sanders for suggesting the comparison between the number of vouchers per square mile and the number of species per square mile (he was the source for the estimated number of species per square mile), David Baxter and Dick Moe for the correct number of heteroduplicates and the total number of duplicate vouchers, and Jane Strong for comments on this page.

Go to:

Copyright © 2015 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:
http://tchester.org/plants/analysis/vouchers/index.html
Comments and feedback: Tom Chester
Last update: 2 March 2015