Somebody read my paper!

 

First of all, BugGuide is great and we love your WYSIWYG lists (although I would like them in poster format so I can print them and post them). Your description of the importance of BugGuide is a good starting point to get more funding. I think BugGuide has become an essential resource that needs to be maintained in a more official capacity.  

 

I don't have any beef with BugGuide. What shocked and offended me was the braggadocio and cavalier attitude of GBIF. Note the title of my paper, the Feynman quote at the beginning, and especially the section under "Ray’s Rule of Precision" which outlines my complaint, a description of the larger more general problem, and an attempt at finding a solution (a brilliant section, if I do say so myself).

 

Concerning GBIF's attitude, see this page as an example: (http://www.gbif.org/using-data/summary). I understand that GBIF needs to present itself as an asset in order to maintain funding, but if they want to be seen as a SCIENTIFIC asset, they are going to have to be more self-critical. For example there should be multiple warnings about data quality, advice on how to check data exhaustiveness, and very nice page listing papers critical of GBIF. Even Wikipedia has this: https://en.wikipedia.org/wiki/Criticism_of_Wikipedia.

 

I was using Bugguide (practically no funding, volunteer and amateur driven, adequate warnings on pages) to contrast GBIF. And, it did just as well! So bully for BugGuide.  

 

Concerning photographs (and I touch on this in the email that was appended, which, by the way is pretty well thought out for something written by a slow witted guy), "evidence of a specimen" lies on a spectrum from sight ID to live specimen. A sight ID (a list of dragonflies seen around a pond) is better than nothing (no one visited the pond). A photograph is better than just sight ID (the photograph represents evidence and is available to hypothesis testing). A dead specimen is good, it can give us many morphological features (some are lost in death), but then we want DNA, then RNA, then proteins, behavior, etc. So even the coveted specimen on a pin isn't the absolute best artifact, it's embedded within the spectrum. There are a lot of (ecology) papers that provide lists of taxa (families, morphospecies, species) with no photographs and no mention of where voucher specimens were deposited (presumably there are none), which is the equivalent of sight ID. A photograph is MUCH better than a sight ID list. But now the conversation is moving in to a topic that I'll touch on at ECN this year...

 

Cheers,

 

Mike


On Sun, Jul 10, 2016 at 11:07 PM, Mike Quinn <[log in to unmask]> wrote:
Dear ECN'ers,

I want to start my defense of BugGuide with a slight push back against some of the comments in Ferro and Flick's (2015) seminal paper on "collection bias" and other far reaching topics. (Links to their paper are at the end of my email.)

Their abstract states that BugGuide "performed poorly" based on the metric of the distribution of the 14 Thoracophorus costalis (Staphylinidae) BugGuide records in relation to their reference model of over 4,900 records of the species borrowed from 38 North American insect collections.

Yes, the 14 BG records only partially mimicked the e. U.S. distribution of the staphylinid in question, but to its credit, BG actually outperformed the majority of the collections consulted. I would further point out that BG ranked in the middle of the "second tier" of collections just ahead of both OSUC and TAMU but behind the top tier of "mega N. Amer. collections" such as CNC, MCZ, FMNH, FSCA and INHS.

Indeed, BG had one of the best rankings in terms of its relative low number of specimens (14) but high ranking on the percent model area scale (Table 1).

I suspect some of the reasons BG performed so well was probably due the fairly even, widespread distribution of its many active contributors.

Google Distribution Map of Top BugGuide Contributors & Top 10 States for Photo Submissions
RED - Current locations of BugGuide contributors who have submitted 5,000 or more photographs 
BLUE - Top 10 states with the most contributed photograph

BugGuide Users
  • Contributors: 31616 (counts anyone who has ever submitted content or comment)
  • Editors: 224
  • Experts: 134
  • Total: 31974
  • Active: 1023 (within the past 7 days)
Link to BG user source data (takes more than a few moments to calculate): http://bugguide.net/node/view/11181

Most mid-sized and certainly smaller collections have a fairly significant regional geographic bias which is less of a factor with BG data, though the Rocky Mountain states and western Canada and Alaska are less well surveyed than the rest of the U.S..

It should also be acknowledged that BG is not quite 13 years old, so for BG to perform just below the mega collections is, I think, particularly impressive.

To be absolutely clear, I am in no way implying that BugGuide is better than physical collections. The gold standard building block of entomological science is, and likely forever will be, a bug on a pin (or other appropriate storage method). 

Personally, I look for ways that BG can serve to cross-pollinate with insect collections. In that vein, I am one of the top contributors to BG and have donated almost every single insect that I photographed  (save for some butterflies) to the Texas A&M Insect Collection or to other collections. Significantly, I have also added some amount of taxonomic, identification, distribution, and host information to nearly every single BG species Info page that I contributed photos to.

Over 5,000 insects (mostly beetles) that I photographed over the last 7 yrs from, or donated to, TAMU.
http://bugguide.net/index.php?q=search&keys=tamuic&search=Search

In this win-win situation, A&M is happy to receive my spmns, I'm happy to have them curated in perpetuity to a very high standard, and the entomological and naturalist communities benefit from having access to both specimens, photographs and associated entomological data of many taxa. Each summer, the contributions to BugGuide continue to grow.

BugGuide Usage Statistics: Images Submitted / Month / Year, Users / Month, Page Views / Month, and Hits / Month

(The first graph suggests we might have hit "Peak BugGuide" in 2014 given that the 2015 photo submission numbers were lower, but 2016 is on track to exceed the 2014 submissions.)

Taken together, the above data attests to the fact that BugGuide has a strong "bottom-up" component as well as a very healthy group of active experts overseeing identifications and other aspects. I suspect that the large number of active users and photos, distribution maps, linked references, and associated biological information all serve to contribute to BugGuide's high ranking in North American taxonomic searches, e.g.


One other "collection bias" Ferro and Flick (2015) discuss in addition to geographic bias is taxonomic bias in terms of different collections having different taxonomic strengths and weaknesses. BugGuide has fairly strong coverage of the most diverse orders with greater representation than the relative diversity of large insects such as walkingsticks, mantids, and roaches. Some orders of small insects, such as thrips and fleas, are underrepresented, but it should be emphasized that BugGuide performed well, depending on one's perspective, in terms of the the small brown staphylinid featured in Ferro and Flick (2015) which is only 2-3 mm in length. 

Orders with more than 1,000 Images
Lepidoptera 268883
Coleoptera 183045
Diptera 140142
Hymenoptera 132056
Hemiptera (sl) 92293
Orthoptera 37077
Odonata 32077
Neuroptera 6117
Trichoptera 5299
Ephemeroptera 4528
Psocodea 3435
Plecoptera 3206
Mantodea 3058
Blattodea 2822
Megaloptera 1525
Phasmida 1253
Mecoptera 1150

Beyond all BugGuide's impressive distribution, usage, and diversity statistics, BG thumbnail photos can also be used in another important way. I may be the only person doing this, but BG thumbnail photos can be simply cut-and-pasted into a WYSIWYG HTML editor. (WYSIWYG = What You See Is What You Get.) This allows for a powerful display of select taxa and specific images. Displaying images in this way can create a tremendous teaching tool, both for the general public and for the professional entomologist outside of their area of expertise. 

Say you live in Oklahoma (or nearby) and you found a strange beetle, but you don't know anything in terms of coleopteran taxonomy which would facilitate narrowing down the options to determine what the beetle might be. The vast majority of the general public probably only vaguely recognizes a few beetle groups such as some scarabs, ladybugs, fireflies (when flashing), longhorn beetles and weevils. Even experienced naturalists can probably only recognize a few more families, but relatively few beetles to the species level.

So here I've constructed a unique page with representative photographs of over 1,200 beetle species, somewhere around a third of the known and a fifth of the projected number of beetles for Oklahoma. To my knowledge, it's never been possible before to see such a large percentage of a state's beetles so easily. The thumbnail photos are 3 cm square (at least on my browser) so most of the beetles are already being viewed at 1 to 8 x magnification without clicking on the thumbnail to enlarge the image further.

Oklahoma Beetle Biota
http://texasento.net/OK_Beetles.html

This compilation of photographs is from surveys I, and others, conducted in and around Oklahoma. It's backed up, and augmented by, two important Oklahoma beetle lists. One is an annotated list from the Oklahoma State University, which has the largest entomological collection in the state. The second list was compiled from material collected by Karl H. Stephan, one of North America's greatest beetle collectors.

Checklist  of the Coleoptera of Oklahoma - K.C. Emerson Entomology Museum (OSEC) 
Department of Entomology and Plant Pathology, Oklahoma State University, Stillwater, OK
http://entoplp.okstate.edu/museum/coleoptera.htm

List of Coleoptera Collected in Latimer County, Oklahoma by Karl Heinz Stephan (2002)
http://bugguide.net/node/view/899478

(OSU provides similar annotated state lists for most insect orders and even for spiders. I'm not aware of any other state providing similar lists for so many orders.)

In summary, I want to recap and add to some of the points above, but also be self critical and acknowledge some of the shortcomings of BugGuide at well.

Strengths of BugGuide:

Was only started in 2004 and already has over a million photographs.
Extremely high usage statistics, e.g. unique visitors, page views and hits per month: http://texasento.net/BugGuide_Statistics_2016.pdf
Each record is tied directly to a photograph vs. most other aggregator sites.
Can easily search data by any combination of taxonomic rank, state, county and/or month.
Photo contributors are fairly dispersed across populated areas of N. Amer. which cuts down on "collection bias" in terms of geography.
Photo contributors are generally interested in a wide range of taxonomic groups which cuts down on "collection bias" in terms of taxonomic emphasis.
Photo contributors are generally available via email for follow up queries.
Taxonomy may be more current than in a physical collection and easier to keep in alphabetical order.
Can accommodate records (photographs) of large sized insects such as saturniid moths with the same ease as accommodating photos of the smallest beetles.
Accessible 24/7 by multiple users worldwide simultaneously, even from handheld devices while out in the field (unless in very remote location).
Thumbnails can be cut-and-pasted directly into a WYSIWYG HTML editor, e.g. http://texasento.net/OK_Beetles.html
If well cropped, the 3 cm sq. thumbnail photo is often sufficiently large to ID most insects, most of which are less than 1 cm long.
In addition to a photo and taxon, BugGuide often has more supplemental information than most other aggregator sites, e.g. http://bugguide.net/node/view/1250146 vs: http://bit.ly/29f7xF6
Photos and associated data are supplied by volunteers so cost is much less than other aggregator sites, (though BG could use more funding to update their software).
Bright colors and natural positions of photographed arthropods can be aesthetically pleasing.

Shortcoming of BugGuide:

As there are so many "cooks" (in terms of contributors and editors), the "broth" can come out a bit muddled...
Often the arthropod isn't collected after being photographed, but personally, I collect nearly everything I shoot, e.g. http://bit.ly/29prYlu
Photographs can be of mixed quality (not that every museum spmn is an intact a thing of beauty!).
Many insects need to be dissected to be ID'ed to the species level, though this problem exists for yet-to-be dissected museum spmns as well.
BG doesn't accept collection dates prior to Jan. 1, 1970 without a workaround (this is a personal pet peeve of mine).
Range limited to N. Amer., north of Mexico, but many collections, particularly smaller ones, are also similarly limited.
General lack of ventral or other non-dorsal views.
Finer morphological details, such as tarsal formulas, often not visible except in the highest quality (stacked) photos.
Many of the dets are somewhat tentative, though not ever museum spmn has a det label and not every det label is actually correct.
Mapping is at the state level instead of the county or point level, thought the county is listed along with the more specific locality.
Can't extract DNA from a photo, though not every collection has an active DNA extraction component.


Michael L. Ferro and Andrew J. Flick. 2015. “Collection Bias” and the Importance of Natural History Collections in Species Habitat Modeling: A Case Study Using Thoracophorus costalis Erichson (Coleoptera: Staphylinidae: Osoriinae), with a Critique of GBIF.org. The Coleopterists Bulletin 69(3): 415-425.


Thanks for considering these points,

Mike Quinn, Austin
________________
Texas Entomology
http://texasento.net


Related previous post to the ECN list:


On Wed, Oct 7, 2015 at 10:17 AM, Mike Ferro <[log in to unmask]> wrote:

Recent research (Ferro and Flick 2015) made me think long and hard (perhaps too long and hard) about museums and the stuff that's in them. I wrote a long, rambling essay on the subject and have pasted it below. It's also available on my blog The eel of Science


Cheers, 

Mike




What is the Best "Collecting" Strategy for an Arthropod Museum?

 

A species consists of a population made up of individuals. Much information about a species can be found in one specimen (especially morphology). However, information about a population (=variation) requires many specimens that exhibit the ranges of morphological and genetic variation, distribution, phenology, etc.

 

Considering that resources are limited, what's a curator to do? Maximize number of species for morphological vouchers, or maximize specimens of a particular species to create a robust representation of the population?

 

 

Short answer:

Do both! Get pretty much everything.

 

 

Long answer 1:

When trying to identify an unknown, even a single identified specimen can be incredibly helpful. Endeavors that benefit greatly from even a single specimen include classes, key creation, field guides, voucher collections (physical and photographic), checklists, etc. So maximizing species has benefits.

 

Most arthropod collections have geographic and taxonomic strengths/biases (see this brilliant paper: Ferro and Flick 2015), that is, they are often dominated by local specimens. (How overlap between "museum range" and species range affects our understanding of a species' distribution is an interesting idea, but not for this essay.) Retaining a large number of specimens of local species may add greatly to the overall understanding of a species' population, after all, who else should be collecting these beasts? So maximizing specimens of species has benefits.

 

Perhaps the best way to answer the question is to consider its inverse: What shouldn't a curator do? Which specimens are of the least value?

 

(Generally speaking!) The specimen that carries the most morphological information is the singleton. Every other specimen of that species provides less and less morphological information, they only contribute information on variation. The specimens that provide the most population information are the outliers (geographic, temporal, etc.) because they help define extremes. Therefore, the specimens that carry the least information represent a species, life stage, time, AND place that is already represented by another specimen in the collection.

 

Specimens of the same species collected at the same time and place are often called a "series", usually defined as 5, 10, or 20 individuals. A series is an important tool for the arthropod collection. It assures that plenty specimens are collected to account for breakage, individual variation, sexes, cryptic species, and abnormal abundance of rare species. But a series can go too far (this is less true for alcoholic preservation where 50 specimens may take up no more room than one). Maybe pointing 100+ ptillids from a single litter sample, or 1000+ throscids from an emergence study is unnecessary (the author is guilty of both).

 

Perhaps the best strategy for a collection manager is to maximize both: 1) the number of species; AND 2) specimens of a particular species, but work to keep redundancy of those specimens below a maximum threshold, and therefore not squander resources on "lower value" specimens.

 

 

Long Answer 2:

 

Not long ago museums had exclusive (more or less) resources such as museum workers, specimens, literature, etc. A user had to travel to the collection to use the resource. And, even though it seems obvious, the collection either had the resource or it didn't (getting the resource, even a paper, took time and money).

 

Digitization of information (and all that that entails) has changed the landscape of cost and access to materials. For example, to peruse The Canadian Entomologist a reader no longer needs to travel to a museum or library—and to possess The Canadian Entomologist a collection no longer needs to have actual copies. It makes little sense to purchase Volume 1 (1869) when it is available for free online. (I'm not saying don't expand your library, just make wise decisions.)

 

For collections a "resource spectrum" has developed. One end consists of resources that are fully digitizeable and can be created, duplicated, and/or passed around with ease including specimen label data. The other end consists of those things which cannot be digitized: the smell of naphthalene, a rousing debate over lunch about whether termites are cockroaches, and above all, physical specimens.

 

Digital "shadows" of specimens can be created, label data, photographs, DNA, etc., but an actual specimen cannot. The cost of obtaining specimens will almost certainly increase with time. Paperwork associated with obtaining, transporting, depositing, and holding a specimen has increased greatly. (These delusional handicaps placed on researchers and institutions are an important and too often overlooked cost/impediment to collections and research, but that's a topic for a different essay.) The point is, from the standpoint of a multigenerational collection, now is probably the best (cheapest) time to obtain specimens (and other non-digitizeable resources).

 

As was mentioned above, even a single specimen provides a lot of information about a species. Additionally, a collection either has a specimen, or it doesn't. Borrowing a specimen from another institution costs time and money. (Enormous amounts compared the first edition of The Canadian Entomologist. Cost to download the first edition of The Canadian Entomologist = $0.00. Cost to mail a package = $10.00. Therefore mailing a specimen is infinity percent more costly than downloading a free paper. That's just math.)

 

In this scenario, increasing species diversity is the best strategy for small or extremely resource-limited collections. Population level information, such as distribution, phenology, etc. can be supplemented through other online specimen-level databases (but beware! see paper cited above).

 

 

My overall proposal is that museums should ACTIVELY work to increase their taxonomic holdings AND actively work to help increase the taxonomic holdings of other museums.

 

Now is the cheapest time (cost will only increase in the future), and actual specimens have clearly emerged as the highest "value" artifacts in a collection (because they are non-digitizeable). It is not inappropriate to imagine a future where all midsized and larger museums have one or more representatives of every order or family of insect, arachnid, etc. known on the continent, or even representatives of every genus (or species of the smaller orders: the 13 smallest orders in NA north of Mexico have a total of 511 spp. Five specimens of each species = 2,555 specimens, or 0.2555% of a million specimen collection.)   

 

Something to mull over…



--
Michael L. Ferro
Collection Manager, Clemson University Arthropod Collection
Dept. of Agricultural and Environmental Sciences
MAIL: 277 Poole Agricultural Center
OFFICE: 307 Long Hall
Clemson University, Clemson, SC 29634-0310
[log in to unmask] (preferred)
[log in to unmask]
https://sites.google.com/site/clemsonarthropodcollection/




--
Michael L. Ferro
Collection Manager, Clemson University Arthropod Collection (CUAC)
Dept. of Agricultural and Environmental Sciences
MAIL: 277 Poole Agricultural Center
OFFICE: 307 Long Hall
Clemson University, Clemson, SC 29634-0310
Subject Editor: The Coleopterists Bulletin; Insecta Mundi