Dragonflies and damselflies (Odonata) are highly skilled aerial acrobats, due largely to the intricacies of their most prominent features: their wings. These structures are both functional and beautiful: they provide extreme maneuverability and durability in a lightweight package, and they can be marked with bright color patterns and even iridescence! The structure and appearance of these wings provides a rich source of information about the evolutionary history, aerodynamic constraints, and behavior of a species.
The Targeted Odonata Wing Digitization (TOWD) Project is part of an NSF-funded, multi-institutional effort to develop ODOMATIC, software for automatically identifying Odonata from images. Initiated in 2016, the TOWD Project aims to digitize the world’s dragonfly and damselfly species, starting with the approx. 466 North American species. We are working to produce a high-resolution imagery of these species (including multiple males and females from each), partnering with several odonate collections, and building the most complete dataset of odonate wings ever compiled. Behind the scenes, we use computer vision to pull phenotypic information from our wings, which can be used for identification and comparative study.
The TOWD dataset will be used for two main purposes:
The TOWD Project’s imagery dataset will be made available to the public on OdonataCentral via the CyVerse cyberinfrastructure, once the project is complete. Until then, you can view our progress below.
We designed an inexpensive digitization setup that uses a commercially-available desktop scanner (connected to a computer) and custom-build paper ‘frame’ placed on the scanner glass to hold things in place and standardize our images. The image below is an example of the final product: one pair of wings is excised and scanned separately from the rest of the body, which is also scanned along with labels and a color standard and scale. Each item gets it’s own ‘window’ in the frame and everything is imaged at once in one high-resolution scan.
Specimen scans are uploaded to our dataset on Cyverse’s BisQue - a powerful imaging platform that allows us to view them from a web browser or work with them from the backend via scripts using BisQue’s API. It’s at this point that one of the most exciting parts of the project takes place behind the scenes, governed by scripts. Our specimen inventory is updated periodically by a script that reads barcodes and labels from the latest scans and checks them against collection databases to get species names and locality information associated with them. Another script detects the wings in each image and automatically measures morphometric properties of each wing (area, length, width, etc.), and also calculates information about its appearance and texture. We can these use these data to make comparisons among species or to update our ODOMATIC species recognition model, using machine learning.
This material is based upon work supported by the National Science Foundation under Grant No. 1564386. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.
Our goal is to image at least 10 males and 10 females from each of the 466 dragonfly and damselfly species in Canada and the US. The progress of this ambitious endeavor (as of 13 December 2018) is illustrated in two forms below.
|Current||Goal||% of goal|
|Individual odonates scanned*||4,073||9,320||43.7%|
|Species begun (1+ scan)||360||466||77.3%|
|Halfway-completed species (10+ scans)||210||466||45.1%|
|*doesn't include "extra" scans above & beyond our goal
**completed species = 10 ♂ + 10 ♀
Jessica is an Associate Professor at Rutgers University-Newark and heads the Insect Systematics (IS) Lab there, where she and her students explore the evolutionary history of Odonata (dragonflies and damselflies) as well as Dictyoptera (cockroaches, termites, and mantises). She leads the TOWD digitization and research efforts at Rutgers.
John is an odonatologist and the Director of Museum Research & Collections at the Alabama Museum of Natural History in Tuscaloosa. He leads the TOWD digitization efforts at ALMNH.
Gareth is an Associate Professor at the New Jersey Institute of Technology and plays an advisory role in the TOWD Project.
Will is a graduate of the IS Lab and is now an NSF Postdoctoral Fellow at the University of Tennessee in Knoxville. The TOWD project is an extension of his dissertation work, in which he digitized a small set of Odonata and developed ODOMATIC. He manages the data workflow for TOWD, as well as this website.
Kim is an Assistant Professor at Rutgers University-New Brunswick and plays an advisory role in the TOWD Project.
Mike is a retired Professor from Rutgers University-New Brunswick and plays an advisory role in the TOWD Project.
Also a graduate of the IS Lab, Melissa worked as a postdoc under the TOWD Project in 2017. There she helped get the Project up and running – developing our digitization protocol, beginning our effort to scan Odonata from several collections with a team of undergraduate students, and much more! She continues to collaborate with us from her new/old home in Bogotá, Colombia.
Dirk is our current TOWD postdoc at the IS Lab. He manages a team of undergraduate students at Rutgers, continuing our scanning effort there. Dirk is also developing additional odonate datasets, seeking to answer questions about flight strategies among dragonflies.
Our work couldn’t be done without the dragonfly and damselfly specimens to digitize. We’ve partnered with several Odonata collections in North America. We’d like to thank our partners for allowing us to work with their collections!
Gassmann, D, J Ware, W Kuhn, JC Abbott, M Sanchez (2018) The Targeted Odonata Wing Digitization (TOWD) project: creating a unique resource for studying wing evolution in dragonflies (Odonata). EntSoc, Vancouver, Canada. Tuesday Nov. 13 10:10 AM (SysEB, Evolution & Diversity of Odonata and Polyneoptera)
Kuhn, WR, JL Ware, JC Abbott, M Sanchez-Herrera, D Gassmann (2017) Automatic dragonfly identification with Odomatic and the Targeted Odonata Wing Digitization (TOWD) project. Int. Congress Odonatology, Cambridge, UK.
Kuhn, WR, G Russell, J Ware (2016) Ode-omatic ID: A system for automatically identifying dragonflies and damselflies from wings. XXV Int’l Congress Entomology, Orlando, FL.
Elmonier, A, D Gassmann, WR Kuhn, J Ware (2018) The relationship between wing color pattern, wing area, and wing weight in Celithemis dragonflies based on the TOWD image database. 10th Annual GS-LSAMP Research Conference. pdf
The behind-the-scene scripts that run the data workflow of this project are publicly available on GitHub.