Besides the fairly obvious fact that this is a web-based dichotomous key, this key differs from those in Needham and Westfall (1955) in several important ways. First, Needham and Westfall more or less attempt to follow phylogenetic lines, which means, for example, that the two-species family Petaluridae is reached as rapidly as the speciose family Libellulidae. My key follows phylogenetic lines only to the extent that key characters do. Second, Needham and Westfall often preferentially split off distinctive taxa, which slows things down for most taxa. For example, within the Libellulinae, Nannothemis keys out immediately, with Pseudoleon close behind, but Erythemis takes 15 or 16 steps to reach. I decided to construct a key that allows the user to identify specimens in all genera, regardless of distinctiveness, as rapidly as possible. The variability in the number of required steps per taxon reflects another issue with the Needham & Westfall keys, namely the occasional odd choices for early, and therefore disproportionately important, couplet characters. For example, an early couplet in the Libellulidae is vein M2 wavy (sometimes slightly so) vs. M2 not wavy. Not only is it not clear how to distinguish slightly wavy from not wavy, but these two dichotomous branches reconnect further down the key, which indicates that numerous genera in fact may or may not have wavy M2. Obviously, all keys have to contend with intraspecific variability; I have tried to minimize the reliance on characters that can potentially vary across a given couplet.

Because Needham and Westfall provided almost all the information for my key, the two should be equally reliable, except in those cases in which I used information from the textual descriptions that was neither used by Needham & Westfall as a key character nor included in their tables of venation characters. Taking into account taxonomic changes since 1955, it takes an average of 8.95 key steps to identify North American dragonflies to genus using the Needham & Westfall keys (ranging from 3 to 15.5 steps per genus). The fast key averages 6.29 steps per genus (ranging from 4 to 8 steps per genus), almost a 50% increase in efficiency. Eleven genera can be identified faster, by up to 3 steps, with Needham & Westfall's keys; 3 genera take the same number of steps in both keys; 48 genera can be identified faster, by up to 10 steps, with the fast key.

Any comments, suggestions, and questions are welcome:

Alan W. Harvey
Department of Biology - 8042
Georgia Southern University
Statesboro, GA 30460
(912) 681-5784
aharvey@gsaix2.cc.GeorgiaSouthern.edu