Division of Comparative Physiology and Biochemistry, Society for Integrative

and Comparative Biology

Surface‐Skimming Stoneflies and Mayflies: The Taxonomic and Mechanical Diversity of

Two‐Dimensional Aerodynamic Locomotion
Author(s): James H. Marden, Brigid C. O’Donnell, Michael A. Thomas and Jesse Y. Bye
Source: Physiological and Biochemical Zoology: Ecological and Evolutionary Approaches, Vol. 73,
No. 6 (November/December 2000), pp. 751-764
Published by: University of Chicago Press . Sponsored by the Division of Comparative
Physiology and Biochemistry, Society for Integrative and Comparative Biology
Stable URL: http://www.jstor.org/stable/10.1086/318109
Accessed: 10-01-2016 19:12 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact [email protected].

Division of Comparative Physiology and Biochemistry, Society for Integrative and Comparative Biology and
University of Chicago Press are collaborating with JSTOR to digitize, preserve and extend access to Physiological and
Biochemical Zoology: Ecological and Evolutionary Approaches.

http://www.jstor.org

This content downloaded from 128.240.233.146 on Sun, 10 Jan 2016 19:12:41 UTC
All use subject to JSTOR Terms and Conditions
 751

Surface-Skimming Stoneflies and Mayflies: The Taxonomic and

Mechanical Diversity of Two-Dimensional Aerodynamic
Locomotion*

James H. Marden† legs to jump from the water into the air to initiate flapping
Brigid C. O’Donnell flight. Comparisons across these forms of skimming show that
Michael A. Thomas wing-beat amplitude, horizontal velocity, and the verticality of
Jesse Y. Bye aerodynamic force production increase as the body orientation
Department of Biology, 208 Mueller Laboratory, becomes more upright and contact with the water is minimized.
Pennsylvania State University, University Park, Pennsylvania These behaviors illustrate a mechanical pathway by which flying
16802 insects could have evolved from swimming ancestors via a series
of finely graded intermediate stages. The phylogenetic distri-
Accepted 4/12/00 bution of skimming and flight in stoneflies does not indicate
any clear directionality toward either greater or lesser aero-
dynamic abilities; however, the broad and apparently basal phy-
logenetic distribution of skimming taxa supports the hypothesis
ABSTRACT
that the common ancestor of stoneflies was a surface skimmer.
The best supported hypothesis for the evolutionary origin of This may also be true for the common ancestor of stoneflies
insect wings is that they evolved from articulated, leg-derived and mayflies, that is, the first winged insects. We combine these
respiratory structures of aquatic ancestors. However, there are data with fossil evidence to form a synthetic model for the
no fossils of the immediate ancestors of winged insects, and it evolution of flying insects from surface skimmers.
is difficult to imagine how a functional transition from gills to
wings could have occurred. Recent studies of surface-skimming
locomotion in stoneflies and mayflies offer a plausible solution
by showing how rudimentary wings and muscle power can be
Introduction
used to accomplish two-dimensional aerodynamic locomotion
on the surface of water. Here we extend that line of research Flight in animals may have originated in species that jumped
by examining the phylogenetic distribution and mechanistic from elevated perches (the trees-down model; Bock 1986) or,
diversity of surface skimming in stoneflies, along with a limited alternatively, from animals that first used a rudimentary form
examination of mayflies. These investigations reveal both a of aerodynamic locomotion while moving across the ground
broad taxonomic occurrence and a fine gradation of mechan- (the ground-up model; Ostrom 1979). Ground-up hypotheses
ically distinct forms. Distinct forms of wing-flapping surface have traditionally been viewed as implausible and weakly sup-
skimming include (1) stoneflies that flap their wings weakly ported (Rayner 1991); however, new data and analyses have
while maintaining their body in contact with the water and given ground-up hypotheses renewed vigor and interest (Padian
undulating their abdomen laterally in a swimming-like motion, and Chiappe 1998; Shipman 1998; Burgers and Chiappe 1999).
(2) stoneflies that skim while elevating their body above the Recent fossil discoveries have greatly strengthened the evi-
water and maintaining all six legs on the surface, (3) stoneflies dence that birds evolved from small running dinosaurs, whose
and mayflies that skim with only four legs on the water surface, ability to perch on tree limbs evolved after, rather than before,
(4) stoneflies that skim with only their two hind legs on the the ability to fly (Padian and Chiappe 1998; Sereno 1999). There
surface, and (5) stoneflies that, beginning with a series of leg is also new evidence for ground-up origins of insect flight.
motions nearly identical to hind-leg skimmers, use their hind Studies of modern insects (Marden and Kramer 1994, 1995;
Kramer and Marden 1997; Ruffieux et al. 1998) have revealed
* This paper was originally presented at the symposium “Water, Land, and Air: nonflying forms of aerodynamic locomotion, called surface
Unifying Principles in Locomotion,” Fifth International Congress of Compar- skimming, which could have been used before the evolution
ative Physiology and Biochemistry, Calgary, 1999.
†
of aerial flight by ancestral semiaquatic insects possessing ru-
To whom all correspondence should be addressed; e-mail: [email protected].
dimentary wings and thoracic muscles. Surface skimming con-
Physiological and Biochemical Zoology 73(6):751–764. 2000. 䉷 2000 by The sists of planar movement across a water surface, wherein pro-
University of Chicago. All rights reserved. 1522-2152/2000/7306-99143$03.00 pulsion is supplied by aerodynamic thrust, while continuous

This content downloaded from 128.240.233.146 on Sun, 10 Jan 2016 19:12:41 UTC
All use subject to JSTOR Terms and Conditions
752 J. H. Marden, B. C. O’Donnell, M. A. Thomas, and J. Y. Bye

contact with the water removes the need for total aerodynamic Drawings of surface-skimming postures were made from print-
weight support. All components of a surface-skimming flight outs of single frames from the high-speed video sequences.
motor (wings, wing articulations, muscles, and neuromotor Velocities were measured from the subset of records in which
patterns) could have simultaneously undergone gradual selec- the insect’s movement was nearly parallel to the film plane.
tion for improved aerodynamic performance before the evo- Surface-skimming ability is strongly dependent on both the
lution of true flight. By providing a scenario that allows for physical integrity of the insect and characteristics of the water.
incremental improvement and elaboration, this model offers a Stoneflies and mayflies were handled as little as possible to
solution to the riddle of how flight could arise through non- prevent damage to their delicate wings and legs. Our preferred
flying intermediate stages. A particularly exciting possibility is method was to capture insects in the field by inducing them
that surface skimming in modern stoneflies may be a retained to walk into clean, dry collection vials from which they were
ancestral trait, which if correct, indicates that we have a unique shaken onto the surface of our test tank (i.e., no direct han-
opportunity to examine living forms that embody the evolu- dling). Tap water was used for all experiments. We avoided any
tionary transition between nonflying and flying animals. contact between our fingers and the water in the test tank since
The surface-skimming hypothesis for insect flight origins has even trace amounts of oils from human skin disrupt surface
been both lauded (Samways 1996; Thomas 1996; Thomas and tension and impair or eliminate skimming.
Norberg 1996) and denounced (Will 1995). Will used a phy- Many of our video recordings were made at locations away
logenetic approach to address the suggestion that surface skim- from the laboratory where we had no control over ambient
ming might be a retained ancestral trait in modern stoneflies. temperature. In order to compare velocity measurements from
However, at the time of Will’s critique, the presence or absence trials performed at different temperatures, we adjusted all skim-
of skimming behavior was known for only two taxa, one of ming velocities to an estimated value at 20⬚C by using the slope
which was ignored in Will’s analysis. Thus, Will’s phylogenetic of a regression of skimming velocity on temperature (0.8 cm/
analysis lacked even a rudimentary sample of the distribution s/⬚C) that was derived from extensive experiments with Tae-
of character states among the taxa in question and therefore is niopteryx burksi at temperatures from 2⬚ to 23⬚C (Marden and
not likely to be informative. Kramer 1994).
A specific goal of this study is to test the hypothesis presented A phylogenetic tree for Plecoptera was constructed using
by Will (1995) that surface skimming is an apomorphic (i.e., PAUP* 4.0b2 (Swofford 1999) with all 113 characters (67 are
recently derived) trait used by a restricted subset of stonefly parsimony informative) coded by Nelson (1984). All characters
(Plecoptera) taxa and that surface skimming in stoneflies bears were coded as unordered (i.e., nonpolar), thus forming an un-
little or no resemblance to locomotor behaviors used by may- rooted tree since Nelson did not sample any out-group taxa.
flies (Ephemeroptera). Similarities between stoneflies and may- There were no missing data or multistate taxa. Reliability of the
flies are particularly interesting because these two groups have tree and homoplasy were examined by calculating the rescaled-
retained many primitive morphological traits (Matsuda 1970; consistency (RC) and retention (RI) indices, based on a heuristic
Brodsky 1994), and each may occupy a basal position (Boud- search (settings p stepwise addition for the starting tree; ran-
reaux 1979; Hennig 1981) in the radiation of the two main dom addition sequence with 100 replicates; tree bisection-
clades of insects, the Paleoptera and the Neoptera (Kristensen reconnection swapping algorithm). These indices can range from
1991). Traits shared by these two taxa are good candidates for 0 to 1, with higher values indicating a more strongly supported
traits possessed by the common ancestor of all flying insects. tree.

Material and Methods Results

Streams in central Pennsylvania were checked frequently for We made high-speed video recordings of surface locomotion
adult stoneflies during all seasons of the year between 1994 for 22 species of stoneflies from 10 families (Table 1). Species
and 1999. In order to include stonefly taxa that are restricted in eight families (Austroperlidae, Gripopterygidae, Taeniopter-
to the southern hemisphere, streams and rivers in southern ygidae, Capniidae, Leuctridae, Nemouridae, Perlodidae, Chlor-
Australia (Victoria and Tasmania) were sampled during a 3- operlidae) utilized some form of wing-propelled surface skim-
w period of October 1998. Mayfly subimagos were also col- ming to move horizontally across the water surface. Of the
lected during the spring and summer of 1998 from central remaining two families, one (Perlidae) contains species that use
Pennsylvania. only runninglike motions of the legs to propel themselves along
Upon capture, stoneflies and mayflies were kept in an in- the surface of water. Another family (Notonemouridae) is rep-
sulated cooler at approximately 10⬚–15⬚C. Within 24 h, they resented in our sample by a single species that we have not
were dropped onto the surface of a shallow tank of water observed to skim; this species instead initiates flight by jumping
(40 # 50 # 3–4 cm) and filmed from the side at 500 frames/ from the water into the air using a powerful thrust of its hind
s with a high-speed video recording system (Redlake HR 1000). legs. As we show below, the hind-leg motions involved in in-

This content downloaded from 128.240.233.146 on Sun, 10 Jan 2016 19:12:41 UTC
All use subject to JSTOR Terms and Conditions
 Surface-Skimming Stoneflies and Mayflies 753

Table 1: Summary of stoneflies sampled, their family affiliation, body length, type of
skimming, and a qualitative judgement of their flight ability
Body
Family and Species Length (mm) Skimming Mode Flight Ability a
Austroperlidae:
Acruroperla atra ............... 12.9 Swim skim No
Tasmoperla thalia ............. 17.2 Swim skim No
Capniidae:
Paracapnia angulata ......... 6.1 Six-leg skim No
Allocapnia vivipara ........... 5.3 Sail No
Chloroperlidae:
Swelta onkos ................... 6.8 Six-leg skim; jump Agile
Gripopterygidae:
Cardioperla diversa ........... 8.0 Four-leg skim Marginal
Dinotoperla cristinae ......... 8.6 Four-leg skim Marginal
Leptoperla sp. .................. 11.4 Four-leg skim Marginal
Leptoperla beroe ............... 11.7 Four-leg skim Marginal
Rickoperla rugosa ............. 5.1 Four-leg skim Marginal
Illiesoperla mayi ............... 8.6 Four-leg skim Marginal
Leuctridae:
Leuctra hippopus .............. 6.5 Hind-leg skim Agile
Leuctra sibleyi ................. 5.0 Hind-leg skim Agile
Nemouridae:
Ostrocerca albidipennis ....... 4.7 Six-leg skim Agile
Paranemoura perfecta ........ 4.7 Six-leg skim Marginal
Amphinemoura nigritta ...... 5.1 Six-leg skim Agile
Notonemouridae:
Notonemoura lynchi .......... 7.9 Jump Agile
Perlidae:
Paragnetina media ............ 16.4 No Marginal
Perlodidae:
Isoperla sp. ..................... 7.8 Six-leg skim; jump Agile
Isoperla sp. ..................... 12.0 Six-leg skim; jump Agile
Taeniopterygidae:
Strophopteryx fasciata ........ 6.5 Six-leg skim Marginal
Taeniopteryx burksi ........... 7.9 Six-leg skim Marginal
a
No p flightless; marginal p poor maneuverability and not all individuals capable of gaining altitude at
room temperature; and agile p can gain altitude and avoid stationary objects while flying.

itiating these jumps are identical to hind-leg motions involved Swimming and Swimming Skimming
in one form of surface skimming.
A striking feature of the diversity of surface-skimming lo- Stonefly nymphs swim by performing rhythmic lateral undu-
comotion in stoneflies is the fine gradation of mechanically lations of their body axis. Taeniopteryx burksi nymphs (Fig. 1A)
distinct forms. Each of these forms of surface locomotion are are fairly representative of swimming by all types of stonefly
described below in rank order of decreasing contact with water nymphs. At a water temperature of 5⬚C, they undulate their
(note that this ordering is based on mechanical rather than body axis at a frequency of 2–3 Hz, accomplishing a net forward
phylogenetic criteria). Readers are encouraged to access a Web- velocity of 2.4 cm/s (SD p 0.1; N p 3).
based version of Figure 1 that contains hyperlinks to platform- Adult Tasmoperla thalia (Austroperlidae) stoneflies use a
independent QuickTime movies of each of these types of form of surface skimming that combines wing flapping with
locomotion (http://www.bio.psu.edu/People/Faculty/Marden/ body undulations similar to the motion used by stonefly
PBZFig1.html). nymphs during swimming (Fig. 1B, 1C). The long axis of the

This content downloaded from 128.240.233.146 on Sun, 10 Jan 2016 19:12:41 UTC
All use subject to JSTOR Terms and Conditions
754 J. H. Marden, B. C. O’Donnell, M. A. Thomas, and J. Y. Bye

abdomen oscillates laterally at a frequency of 7 Hz through an

arc of approximately Ⳳ15⬚ in relation to the long axis of the
head and thorax (air temperature p 23⬚C). The phase of this
oscillation appears to be independent of wing motions.
The middle (mesothoracic) and hind legs of T. thalia do not
hold the abdomen above the water. The mesothoracic legs are
flexed under the thorax, and the hind legs extend posteriorly
alongside the abdomen (Fig. 1C). Thus, there is extensive con-
tact between the ventral body surface and the water so that
swimming motions may contribute to thrust and maneuver-
ability. It should be noted, however, that swimming motions
probably degrade aerodynamic performance since the leading
edges of the wings are oriented at angles other than perpen-
dicular to the oncoming airflow at certain points of the wing-
stroke cycle (Fig. 1B). Certain insects use a nonoscillatory lateral
flexure of the abdomen as a steering mechanism (Grodnitsky
1999; Dudley 2000), but we know of no other insects in which
the abdomen oscillates laterally during wing flapping.
Tasmoperla thalia flaps its wings at a frequency of 20 Hz
through a stroke amplitude of approximately 90⬚. The wings
clap together and are peeled apart (Weis-Fogh 1973; Ellington
1984) at the start of the downstroke. The forewings slightly
lead the hind wings throughout the stroke cycle. This kinematic
pattern, called “anteromotoric in-phase functional four-
wingedness,” is thought to be the ancestral condition for flying
insects (Grodnitsky 1999). Pronation of the forewings at the
start of the downstroke and suppination at the start of the
upstroke (Fig. 1B, right panel) creates the kinematic asymmetry
between the downstroke and upstroke that is necessary for
producing a net force. Video records show that the tip of the
forewing bends forward during the initial portion of the up-
stroke, which indicates an aerodynamic force acting on the wing
in a direction that is parallel to the longitudinal body axis, that
is, thrust. Skimming velocity of T. thalia averaged 15.6 cm/s,
which is considerably faster than the swimming velocity of
stonefly nymphs but slower than all other types of surface
skimming (Fig. 2).
Another austroperlid stonefly, Acruroperla atra (not depicted
in Fig. 1), skims without body undulations but with a similar
body position and extensive contact between the ventral ab-
domen and the water. This species raises its thorax above the
water by extending the mesothoracic legs anteriorly. However,
the entire tibia of these legs remains on the water surface, and
the extensive water contact creates a large amount of drag.
Thus, like T. thalia, this species is a comparatively slow skimmer
(mean velocity p 22.4 cm/s; air temperature p 20⬚C).

Figure 1. Drawings made from individual frames of video recordings

of different forms of aquatic and semiaquatic locomotion used by
stoneflies and mayflies. All except swimming (series A) depict move-
ment on the surface of water. Samples of video recordings are available
at http://www.bio.psu.edu/People/Faculty/Marden/PBZFig1.html.

This content downloaded from 128.240.233.146 on Sun, 10 Jan 2016 19:12:41 UTC
All use subject to JSTOR Terms and Conditions
 Surface-Skimming Stoneflies and Mayflies 755

species use a limited form of six-leg skimming as a conditional

strategy to remain mobile at cool air temperatures.
Wing-beat kinematics during six-leg skimming are much the
same as described above (“Swimming and Swimming Skim-
ming”) for T. thalia, except that elevation of the body above
the water allows a greater stroke amplitude. The wing stroke
of six-leg skimmers extends approximately 20⬚ below the hor-
izontal, which extends the total stroke amplitude to 110⬚. Wing-
beat frequency of six-leg skimmers reaches frequencies as high
as 45 Hz in Strophoteryx fasciata (Taeniopterygidae) at warm
air temperature (22⬚C). The combination of greater stroke am-
plitude and frequency yields a much greater angular velocity
of the wings, which increases aerodynamic force production
(Lehman and Dickinson 1997). Larger aerodynamic forces and
reduced contact with the water result in higher skimming ve-
locities (mean p 32 cm/s; N p 7 species from six families;
SD p 2.9; Fig. 2).
Flight ability of six-leg skimmers varies widely. At the low end
of the scale is the capniid Paracapnia angulata, which attempts
Figure 2. Mean velocity of stoneflies using different forms of aquatic,
flapping flight when dropped or thrown into the air but even at
semiaquatic, and aerial locomotion. Sample sizes refer to the number
of species; error bars represent the standard error of the grand mean warm air temperatures (25⬚C) can only spiral downward to a
(i.e., variability among means derived from multiple species). Note the spot on the ground almost directly below the point from which
trend of increasing velocity as contact with water decreases. Body size it was dropped. The capniid that accomplishes six-leg skimming
had no significant effect on skimming speed. by sailing rather than flapping (A. vivipara; Fig. 3) is also flight-
less, although it does extend its wings and glide when tossed into
Even at warm temperatures (20⬚–25⬚C), we were unable to the air (Marden and Kramer 1995). Other six-leg skimmers are
elicit flight attempts from any T. thalia individuals (N p 3). marginally flight capable. At warm air temperatures, approxi-
Wing flapping in this species was observed only as an immediate mately 50% of individuals of Taeniopteryx and Strophopteryx
and stereotypical response to contact with water. A. atra did stoneflies can maintain or gradually gain altitude, whereas the
attempt to fly when tossed into the air, but none of the six remainder cannot remain airborne (Marden and Kramer 1994).
individuals tested were able to maintain altitude. Thus, it ap- Certain nemourids, chloroperlids, and perlodids are agile fliers;
pears that these species are not capable of flying. they are capable of steeply angled takeoffs and can turn to evade
fixed and slowly moving objects.
Six-Leg Skimming
Four-Leg Skimming
Species (N p 6) that we have examined in the families Tae-
niopterygidae, Nemouridae, and Capniidae use their legs to We have examined the skimming behavior of six species (Car-
elevate their entire body above the water while they continu- dioperla diversa, Dinotoperla cristinae, Leptoperla beroe, Lepto-
ously flap their wings during surface skimming (Fig. 1D). Be- perla sp., Illiesoperla mayi, and Ricoperla rugosa) of stoneflies
cause these taxa maintain the tarsi of all six legs in continuous in the family Gripopterygidae, the most diverse plecopteran
contact with the water surface, we have termed this form of family in the southern hemisphere. Gripopterygids raise their
locomotion “six-leg skimming.” An additional capniid species forelegs above the water surface (Fig. 1E), which creates a more
that we have examined, Allocapnia vivipara, shows no ability upright body angle and a greater wing-stroke amplitude. Stroke
to flap its wings but also accomplishes six-leg skimming by amplitude of the forewings is nearly 180⬚, whereas the hind
raising its wings in response to wind, thereby sailing across the wings are constrained by the water surface to a stroke amplitude
water surface (Fig. 3; Marden and Kramer 1995). of approximately 110⬚–120⬚. Wing-beat frequency ranges
At cool air temperatures (!15⬚–18⬚C), stoneflies in the families widely among four-leg skimmers, from a low of 23 Hz in Il-
Chloroperlidae and Perlodidae use intermittent forms of six-leg liesoperla to a high of 42 Hz in Leptoperla.
skimming (N p 3 species; we have also observed this behavior The more upright posture of four-leg skimmers creates a
in four additional chloroperlids that we have not identified). wing-stroke angle that orients the net aerodynamic force more
These stoneflies periodically beat their wings through 3–5 cycles vertically. Development of a significant vertical force during
and then coast. At warmer air temperatures, they jump from the four-leg skimming is readily apparent in one of our films, which
water to initiate flight (see “Jumping from Water”). Thus, these shows a skimming gripopterygid stonefly becoming airborne

This content downloaded from 128.240.233.146 on Sun, 10 Jan 2016 19:12:41 UTC
All use subject to JSTOR Terms and Conditions
756 J. H. Marden, B. C. O’Donnell, M. A. Thomas, and J. Y. Bye

Figure 3. Photograph of a short-winged Allocapnia vivipara male. This species does not flap its wings but raises them in response to wind,
thereby accomplishing sailing (Marden and Kramer 1995).

for a few wing-stroke cycles before it stalls and returns to the One possibility is that cerci counteract a tendency to pitch
water surface (http://www.bio.psu.edu/People/Faculty/Marden/ backward. Another possibility is that cerci widely distribute
movies/FourLeg.mov). Presumably, there is partial aerody- downward forces generated by the tip of the abdomen, thereby
namic weight support during normal four-leg skimming, which reducing surface deformation and drag.
should have the beneficial effect of reducing indentation of the
water surface and consequently the size of the bow wave pushed
Hind-Leg Skimming
by the legs, that is, less drag. Velocity of four-leg skimmers
averaged 44 cm/s (SD p 16.6; Fig. 2). Stoneflies in the family Leuctridae (Leuctra hippopus and Leuc-
Our video records of skimming gripopterygids were obtained tra sibleyi) use a form of skimming (Kramer and Marden 1997)
at air temperatures of 20⬚C. In the field at cooler air temper- that we have not observed in any other taxa. At air temperatures
atures (12⬚–18⬚C), gripopterygids were incapable of flying but that are too cool to fly (!13⬚C), Leuctra stoneflies use a ste-
reflexively (i.e., immediately on contact with the water) utilized reotypical series of motions that raise their body into a nearly
surface skimming when dislodged from midstream rocks. They vertical posture. The hind legs are flexed anteriorly to bring
readily reached shore even on fast-moving cold water. At the tarsi forward to a position under the middle of the thorax
warmer air temperatures (18⬚–23⬚C), gripopterygids were mar- (Fig. 4, upper series). Subsequent extension of the hind legs,
ginal fliers, barely able to maintain altitude. However, on a accompanied by the first downstroke of the wings, raises the
warm sunny day (23⬚C), we observed thousands of Dinotoperla body into a nearly vertical posture. This posture is maintained
spp. ascending into the air above the Delatite River (Victoria, during continued wing flapping, in which both the fore- and
Australia). They were air-driven upward by thermal convection hind wings achieve a stroke amplitude that approaches the
that carried the marginally flight-capable stoneflies above the maximal 180⬚ (Fig. 1G). This is not an entirely stable body
tops of the trees and apparently far from the river. This ob- position because there is sometimes excessive forward pitch
servation demonstrates that flight and dispersal ability need not that is controlled by temporarily placing the tarsi of the middle
be tightly coupled, since air currents can greatly augment the legs down on the water. During most wing-beat cycles, Leuctra
mobility of insects that are just barely able to remain airborne. has only its two hind tarsi and the tip of its abdomen in contact
A number of mayfly subimagoes (Ephemeroptera) also use with the water. As noted above for other types of skimming,
four-leg skimming at cool air temperatures (!13⬚C; Fig. 1F). video records of Leuctra show that the forewings bend during
Their forewings achieve a 180⬚ stroke amplitude, and there is the wing stroke, thus indicating a net aerodynamic force acting
no restriction on the stroke amplitude of the hind wings, which on the wing in a direction that is parallel to the longitudinal
in mayflies are much smaller than the forewings. Mayfly skim- body axis. This is a vertically oriented net force when the body
ming features a slight pitching motion of the body during the is oriented in a head-up posture. By minimizing body contact
wing-stroke cycle, which causes the hind legs and the tip of with water and by producing vertical force, hind-leg skimmers
the abdomen to rise off the water during the latter half of the minimize drag on the water surface. Average velocity of Leuctra
downstroke. Only the tarsi of the middle legs remain in contact was 42 cm/s (SD p 1.9), which is not significantly different
with the water throughout the entire wing stroke. from the mean velocity of four-leg skimmers (44 cm/s) but is
It is interesting to note that both gripopterygid stoneflies and significantly higher than all other forms of skimming (Fig. 2).
mayflies have long abdominal cerci that trail behind the body Adult Leuctra stoneflies are active during spring (late April
on the water surface. The consistent association between four- and early May in Pennsylvania) when the weather is highly
leg skimming and long cerci suggests a functional connection. variable. At warm air temperatures, Leuctra are relatively strong

This content downloaded from 128.240.233.146 on Sun, 10 Jan 2016 19:12:41 UTC
All use subject to JSTOR Terms and Conditions
 Surface-Skimming Stoneflies and Mayflies 757

Figure 4. Sequence of movements and body positions (progressing from left to right) during the initiation of hind-leg skimming (Fig. 1G) by
stoneflies in the family Leuctridae (upper series) and takeoff from the water (Fig. 1H ) by stoneflies in the family Notonemouridae (lower series).
Note the similarity in the way the hind limbs are used and positioned during these two forms of locomotion.

fliers, whereas at cool air temperatures, we have observed large and Kramer 1994, 1995) that surface skimming is a plesio-
numbers of L. sibleyi using hind-leg skimming to move about morphic (i.e., basal) trait. Will used morphological character
on the surface of a stream. Thus, it appears that Leuctra stone- data from Nelson (1984) to construct a phylogenetic tree of
flies use surface skimming to remain mobile over a broad range Plecoptera families. This tree was presented as if it were a rooted
of ambient temperatures. phylogeny; however, Nelson’s data set contains no analysis of
taxa other than Plecoptera. Thus, the out group labeled OG in
Will’s figure (1995, Fig. 1) does not exist, and the rooting of
Jumping from Water the tree remains uncertain. Nevertheless, it is possible to use
Stoneflies in the families Notonemouridae, Chloroperlidae, and Nelson’s data to construct an unrooted phylogeny and thereby
Perlodidae use their hind legs to jump from water into the air, test Will’s hypothesis that surface skimming occurs only in an
thereby initiating aerial flight (Fig. 1H). The series of leg mo- isolated place on the tree, which would indicate that it is most
tions used to perform this jump are strikingly similar to those likely an apomorphy.
used by Leuctra stoneflies to initiate hind-leg skimming. Jump- Our phylogenetic analysis of Nelson’s morphological char-
ers flex their hind legs, thus bringing the tarsi forward to a acter data yielded 12 equally parsimonious trees with 153 steps
position under the middle of the thorax (Fig. 4, lower series). (for each tree, RC p 0.61 and RI p 0.83). A strict consensus
Rapid extension of the hind legs accompanied by the first of these trees is shown in Figure 5, upper panel. Onto this tree,
downstroke of the wings launches the insect into the air. Flight we have mapped the occurrence of particular types of surface
is then maintained by continued wing flapping while the body skimming. Skimming is universal among taxa belonging to the
remains oriented at an angle similar to that at takeoff (Fig. 4, superfamily Nemouroidea and is present in all of the taxa ex-
lower series, far right panel). Mean horizontal velocity of flying amined to date within the clade that contains the superfamilies
stoneflies exceeded that of all forms of surface skimming (Fig. Eusthenioidea and Gripopterygoidea (Fig. 5, lower panel). A
2; mean p 63 cm/s, SD p 15; N p 7 species). very limited form of skimming (intermittent rather than con-
tinuous flapping) is used by certain taxa in the superfamily
Perloidea, although stoneflies in this group are much more
Phylogenetic Distribution of Skimming in Plecoptera
inclined to jump from the water and fly than they are to skim.
Will (1995) began to examine the question of how surface The only taxon sampled to date that shows no use of winged
skimming and flying are distributed phylogenetically among locomotion on the water surface (the Perlidae) is in this clade.
stoneflies with the aim of evaluating our hypothesis (Marden We have not yet examined any species in the clade containing

This content downloaded from 128.240.233.146 on Sun, 10 Jan 2016 19:12:41 UTC
All use subject to JSTOR Terms and Conditions
758 J. H. Marden, B. C. O’Donnell, M. A. Thomas, and J. Y. Bye

Figure 5. Maximum parsimony tree (unrooted; rescaled-consistency index [RC] p 0.61 , and retention index [RI] p 0.83 ) constructed using
67 parsimony-informative morphological characters from data in Nelson (1984). Surface-skimming behavior is mapped onto the tree in the
upper panel by using graphic icons from Figure 1. Except for Perlidae, which display no use of winged locomotion on the water surface, and
Scopuridae, which are wingless, taxa without icons have not yet been tested for surface locomotion behavior. The lower panel shows numbers
that indicate the date (millions of years ago) of the first fossil record for each family (Labandeira 1994). Note that there are older fossils (ca.
260 million years ago) of what appear to be taeniopterygids and eustheniids (Illies 1965), so these dates are conservative.

This content downloaded from 128.240.233.146 on Sun, 10 Jan 2016 19:12:41 UTC
All use subject to JSTOR Terms and Conditions
 Surface-Skimming Stoneflies and Mayflies 759

Pteronarcys, Allonarcys, or Pteronarcella, nor have we exam- mer 1995). Together, these data constitute the only empirical
ined any peltoperlids. Scopurids are wingless and therefore not demonstration of a pathway by which flapping flight could have
amenable to analyses of winged locomotion. evolved through a series of intermediate stages. It is worth
noting that certain reptiles (Glasheen and Mcmahon 1996) and
birds (Livezy and Humphrey 1985) are also active on the surface
Discussion
of water. Thus, general features of the pathway that we have
The Diversity of Surface Skimming in Stoneflies proposed for the evolution of insect flight may also be a rea-
sonable mechanical model for the evolution of avian flight.
Alternatively, the water surface may only be a particularly ad-
The stoneflies are the most primitive members of the contem- vantageous setting for insect flight evolution since larger ani-
porary winged Polyneoptera. They have retained a lot of primitive
mals can reach higher running speeds and are less impeded by
structural features, such as thysanuran-like nymphs, the meso- and
the irregular surfaces of terrestrial habitats. Ground-up evo-
metathorax movable relative to each other, the absence of a distal
medial plate in the axillary apparatus, an equal stroke amplitude lution of flight in vertebrates may have occurred through a
by fore- and hind wings, the same yawing flight as in mayflies, and roughly analogous series of stages by terrestrial vertebrates run-
so on. All this makes stoneflies especially interesting. (Brodsky 1994, ning over solid surfaces (this is essentially what has been pro-
p. 116) posed in Burgers and Chiappe [1999]).
To address the second question regarding the evolutionary
As suggested by Brodsky (1994), the behavior and mechanics history of aerodynamic performance in modern stoneflies, we
of stoneflies are potentially highly relevant for understanding have mapped surface-skimming behavior onto a phylogenetic
the origins of insect flight. Despite this potential, little effort tree for Plecoptera (Fig. 5). This figure shows that skimming
has been made to examine how this widespread group of insects is much more widespread among stonefly taxa than has pre-
accomplishes winged locomotion. Our recent studies (Marden viously been assumed. Rather than being an apomorphy (Will
and Kramer 1994, 1995; Kramer and Marden 1997) and the 1995), skimming is ubiquitous among the Gripopterygoidea
data presented here show that stoneflies utilize a wide variety and Nemouroidea superfamilies. Families within these two
of mechanically distinct forms of surface skimming. These data clades are among the oldest Plecoptera, with fossil records dat-
can now be used to address two distinct questions: (1) Is there ing back as far as 248 million years (Fig. 5, lower panel; La-
a likely mechanical pathway for the gradual evolution of flying bandeira 1994). Families within the Eusthenioidea (Fig. 5, lower
from skimming, and (2) What is the evolutionary status (an- panel) are equally old, but unfortunately, we did not find any
cestral or derived) of the skimming behavior of modern adults to examine for skimming behavior during our 3 w of
stoneflies? fieldwork in Victoria and Tasmania (note, however, that as we
The answer to the first of these questions is that there is a were preparing the final draft of this paper, we found a eus-
very clear stepwise pathway that could have been used to thenioid stonefly in Chile, Diamphipnopsis samali, that uses its
achieve the transition from aquatic to aerial locomotion (Fig. wings to row along the water surface; this behavior will be
1). At the base of this hypothetical pathway is skimming using described in a subsequent publication). The Eusthenioidea have
a horizontal body posture, a mixture of swimming and flying traditionally been assumed to be the basal lineage of stoneflies
motions, and extensive contact between the body and the water (Illies 1965); however, our tree suggests that the Gripoptery-
surface (Fig. 1B, 1C). A key innovation from this stage would goidea and Eusthenioidea are a monophyletic group that con-
have been using the legs to elevate the body above the water, tains, in the case of austroperlids, flightless stoneflies that com-
thereby reducing drag (Fig. 1D). A next step would be to raise bine skimming with swimming (Fig. 1B, 1C). In general, the
the forelegs off the water surface and elevate the anterior por- widespread occurrence of skimming among the oldest families
tion of the body, thereby further reducing drag on the water, of Plecoptera supports the hypothesis that skimming is a re-
enhancing wing stroke amplitude, and orienting the net aero- tained ancestral trait. We present elsewhere a rooted molecular
dynamic force more vertically (Fig. 1E, 1F). Raising the middle phylogeny of Plecoptera based on 18S rRNA sequence data
legs off the water allows further enhancement of wing-stroke (Thomas et al., in press) that indicates a well-supported basal
amplitude and drag reduction (Fig. 1G). Finally, elaboration of position of the superfamily Nemouroidea in which all of the
the leg motions used to elevate the body into a head-up skim- species that we have sampled (N p 6 species) are six-leg skim-
ming posture may have been used to propel the insect into the mers. This does not imply that the first stoneflies were flightless,
air, thereby accomplishing true powered flight (Figs. 1H, 4). since most of the stonefly species that we have examined, in-
We have shown previously (Marden and Kramer 1994) that cluding Nemourids and Taeniopterygids, are capable of flying
stoneflies can accomplish skimming with experimentally re- in addition to surface skimming (Table 1). Thus, the ancestral
duced wing size and muscle power output and that certain condition for Plecoptera was most likely a combination of
extant stoneflies possess and utilize rudimentary wings and skimming and flying. In general, the order Plecoptera is a group
wing motions for surface skimming (Fig. 3; Marden and Kra- of insects that are dual-mode aerodynamic locomotors; they

This content downloaded from 128.240.233.146 on Sun, 10 Jan 2016 19:12:41 UTC
All use subject to JSTOR Terms and Conditions
760 J. H. Marden, B. C. O’Donnell, M. A. Thomas, and J. Y. Bye

use their wings to move in two dimensions across water sur- male oviposition on the water surface, may simply be a three-
faces, and most species are able to fly in three dimensions (albeit dimensional extension of an ancestral mating system that was
quite weakly compared to most types of insects). carried out using four-leg skimming on the surface of water
(Fig. 6). Stoneflies and dragonflies (Odonata) also oviposit on
water and share with mayflies a simple egg chorion that cannot
Surface Skimming in Mayflies withstand desiccation (Hinton 1981). These features may be
the result of convergent evolution, but it is also possible that
Ephemeroptera are widely believed to possess many primitive they are shared ancestral traits. If so, then scramble competition
features relevant to the origins of the pterygotes. (Kingsolver and for access to females and evasion of surface-feeding predators
Koehl 1994, p. 429) may have selected for progressively faster forms of surface skim-
ming (Fig. 2) and ultimately the adoption of aerial mating
Certain mayflies (Ephemeroptera) use four-leg skimming in a systems that minimized contact with the water surface.
manner that is mechanically similar to gripopterygid stoneflies
(Fig. 1E, 1F). However, they do so only in the subimago stage
and only when they are too cold to fly (i.e., air temperatures Evolutionary Origins of Insects and Their Wings
below about 13⬚C). At warmer temperatures, they are strong
fliers, departing immediately from the water surface and show- Proto-wings probably functioned initially as respiratory organs
ing no conspicuous use of four-leg skimming in the field. This in amphibious or aquatic conditions (as do the flattened exites of
is appropriate behavior for insects that are active in seasons other arthropods). (Kukalova-Peck 1991, p. 144)
and habitats where surface-feeding fish are abundant and highly
active and eagerly consume mayflies that linger on the surface. Recent phylogenetic analyses based on both molecular (Fried-
Most of the stoneflies that we have studied make either oc- rich and Tautz 1995; Regier and Schultz 1997; Aguinaldo and
casional or frequent use of surface skimming in the field, but
they emerge earlier in the spring than do most mayflies, when
the water is cold and fish are not yet feeding on the surface.
The hypothesis that skimming behavior is related primarily to
predation, and only secondarily to temperature, is supported
by the finding that a mayfly species in tropical Madagascar has
lost the ability to fly yet moves about on the water surface using
four-leg skimming (Ruffieux et al. 1998; M. Sartori, personal
communication). Male imagoes skim to search for females, and
the females do not progress beyond the subimago stage. No-
tably, this species exists in a river system that lacks surface
feeding fish, which suggests that in the absence of modern
forms of predation, the locomotor and mating behavior of
modern aquatic insects, even at warm temperatures, can readily
collapse from three-dimensional to two-dimensional and still
remain mechanically and ecologically viable.
Surface skimming in mayflies has two other potentially im-
portant evolutionary implications. First, because mayflies and
stoneflies represent relatively basal lineages in each of the two
major branches of winged insects (the Paleoptera and Neoptera;
Kristensen 1991), traits shared by these two groups, such as
skimming and wet-resistant hydrofuge hairs on the wings (Mar-
den and Kramer 1994), may have been retained from the initial
radiation of winged insects. Second, because the forelegs of
four-leg skimmers are held above the water (Fig. 1E, 1F), they Figure 6. Drawings showing a hypothetical progression (in bottom-
are free to be used for functions other than support and lo- to-top order) for locomotion and foreleg size in male mayflies. The
comotion. If ancestral mayflies carried out their mating activ- bottom image (hypothetical) shows six-leg skimming with normal-
ities using four-leg skimming, the forelegs of males might have sized forelegs; the middle image shows four-leg skimming with elon-
gated forelegs used for grasping females encountered on water (as
become elongated and specialized for grasping females before occurs in the species described in Ruffieux et al. [1998]); and the top
mayflies evolved the ability to fly. The mating system of modern image shows a modern species with elongated forelegs used for grasp-
mayflies, which involves both elongated male forelegs and fe- ing females during aerial flight.

This content downloaded from 128.240.233.146 on Sun, 10 Jan 2016 19:12:41 UTC
All use subject to JSTOR Terms and Conditions
 Surface-Skimming Stoneflies and Mayflies 761

Lake 1998; Boore et al. 1998; Winnepenninckx et al. 1998;

Garcia-Machado et al. 1999) and morphological (Strausfeld
1998; Strausfeld et al. 1998) characters support the hypothesis
that insects are a sister clade of the Crustacea. Recent data also
indicate a homology between crustacean gills and insect wings
based on the demonstration that two of the genes involved in
morphogenesis of a particular dorsal exite of the brine shrimp
gill are involved in a similar stage of development in Drosophila
wings (Averof and Cohen 1997; but see Jockusch and Nagy
[1997] and Williams [1999] for notes of caution in interpreting
this apparent homology). Other support for the wings-from-
gills hypothesis comes from the observation that Drosophila
carrying mutant forms of the ultrabithorax and abdominal-a
homeotic genes develop wing primordia on the first seven ab-
dominal segments (Carroll et al. 1995), in the same location
that both fossil mayflies and modern mayfly nymphs have gill
plates. These gill plates have long been argued to be serial
homologs of wings, the thoracic equivalents of which were the
evolutionary and mechanical precursors to wings (Kukalova-
Peck 1978, 1983, 1987, 1991). Thus, it appears that modern
homeotic regulation represses the initiation of wing/gill devel-
opment on all but the meso- and metathorax of extant pter-
ygotes (Carroll et al. 1995). The wings-from-gills model is also
supported by the finding that wings carry sensilla and che-
moreceptors that are present on arthropod appendages (i.e.,
the leg exites that became gills) but not on the thoracic body
wall (Dickinson et al. 1997). Together with fossil data, these
studies represent a growing body of evidence that insects and
their wings evolved from a crustacean-like ancestor and its
moveable leg-derived gills.
A fundamental difficulty for the wings-from-gills hypothesis
has been the implausibility of gills being maintained during the
terrestrial and presumably arboreal stage that preceded flying
insects. This argument is based largely on the assumption that
terrestriality is the primitive condition for insects, dating back
at least as far as a terrestrial Hexapoda-Myriapoda stem group Figure 7. Fossil insects with rudimentary wings and abdominal struc-
(Pritchard et al. 1993), but this interpretation has been rendered tures that may have functioned as gills during activity on the water
obsolete by the present understanding that myriapods and surface. The top two panels (a, b) show variations in wing and body
size that are thought to represent ontogenetic stages (i.e., instars) within
hexapods are not closely related and that hexapods are the sister
species. These fossils are representatives of the stem groups that gave
clade of crustaceans. A crustacean-like root for insects makes rise to the modern orders Ephemeroptera (a), Hemiptera (b), and
it much more plausible that apterygote insects lost gills (perhaps Plecoptera (c). Panel d belongs to the order Megasecoptera, part of
independently, since apterygotes do not appear to be a mono- the extinct superorder Paleodictyoptera. Drawings are reproduced from
phyletic group) as they invaded land, thereby diverging from Kukalova-Peck (1978, 1991).
a gill-bearing stem that ultimately gave rise to pterygote insects
(Toms 1984). A prediction that emerges from this model is did in fact serve as gills, or even if these insects were associated
that the fossil record should contain insects that possess both with the water surface, this is the first attempt to offer a syn-
wings and gills, that is, early stages in the evolution of surface thetic explanation for the common co-occurrence in fossils of
skimmers. Indeed, such fossils from the upper Carboniferous rudimentary wings and abdominal structures that look like gills.
and lower Permian are abundant and taxonomically diverse Surface skimming provides a behavioral and environmental
(Fig. 7), including examples from the stem groups that led to context in which gills and rudimentary wings could have co-
modern Ephemeroptera, Hemiptera, and Plecoptera along with existed during a time when tracheae were perhaps gradually
the extinct order Megasecoptera (Fig. 7). Although it cannot invaginating and becoming elaborated, ultimately replacing an
be determined that the abdominal structures of these fossils ancestral blood-based system of gas exchange. Such a scenario

This content downloaded from 128.240.233.146 on Sun, 10 Jan 2016 19:12:41 UTC
All use subject to JSTOR Terms and Conditions
762 J. H. Marden, B. C. O’Donnell, M. A. Thomas, and J. Y. Bye

also refutes the hypothesis that tracheae-bearing modern insects of the anterior part of the body would reduce the ability of the
are necessarily secondarily aquatic (Pritchard et al. 1993; Grod- thoracic gills to contribute to swimming but would have freed
nitsky 1999), since tracheal respiration may have originated on them to become specialized for other forms of locomotion.
the water surface in a gill-bearing lineage. This may have occurred initially as rowing in a manner similar
to what occurs in Diamphipnopsis stoneflies and in a small
group of marine chironomid flies that apparently use their
A Synthetic Model for the Evolution of Flying Insects from greatly modified wings to row along the ocean surface (Cheng
Aquatic Ancestors and Hashimoto 1978; this particular case is undoubtedly a
recent loss of normal Dipteran wing function but nonetheless
demonstrates a mechanically feasible role for early protowings).
Most of the literature [about insect flight origins] hasn’t reached Alternatively, small gill plates may have been useful primarily
the first base of applying functional shift in the first place, not to as sails (Fig. 3), as occurs in modern Allocapnia vivipara stone-
mention the later stations of substituting direct evidence for verbal
flies that raise their wings in response to wind currents (Marden
speculation. Most reconstructions are still trying to explain the in-
and Kramer 1995). Increasing the frequency of wing-raising
cipient stages of insect wings as somehow involved in airborne
performance from the start—not for flapping flight, of course, but and lowering motions would constitute crude flapping and
still for some aspect of motion aloft rather than, as Darwin’s prin- some degree of independence from wind.
ciple would suggest, for some quite different function. (Gould 1985, Biotic factors may also have been important contributors to
p. 21) the evolution of activity on the water surface. Terrestrial Car-
boniferous habitats featured great forests of wind-pollinated
We have met the challenge laid out by Gould’s (1985) critique plants, whose pollen, seed pods, and leaves presumably col-
by demonstrating how real insects can use rudimentary flight lected in large quantities on the surface of freshwater swamps
motor function for something other than motion aloft. How- and lakes. Floating pollen and seeds accumulate along the
ever, a robust model for the evolution of insect flight from downwind shore of bodies of water, where they would have
surface skimming must consider ecological factors in order to been abundant, easily collected, and nutritious food sources
understand how natural selection could have favored surface for surface-dwelling insects. The hypothesis that surface-based
activity by formerly aquatic insects and how surface skimming pollen feeding may have been an important niche for early
could give way to flying. Rather than emphasize any single insects is consistent with the large percentage (50%) of Car-
factor, we prefer to argue that many environmental factors boniferous fossil insects that possessed sucking or semisucking
could have acted in concert. One potentially important abiotic mouthparts. For example, haustellate paledictyopteroids are
factor is the much greater availability of oxygen in air than in thought to have imbibed mostly “spores” or “juices” (Kuka-
water (Schmidt-Nielsen 1990), especially during the Carbon- lova-Peck 1991). Scramble competition for floating foods as
iferous, when atmospheric oxygen may have been elevated well as access to mates and avoidance of predators that evolved
(Graham et al. 1995; Dudley 1998) and organic decomposition to exploit surface-active insects would eventually have favored
in freshwater swamps is likely to have greatly depleted dissolved faster and more mobile surface skimmers, that is, a progression
oxygen. through the mechanical stages and velocities shown in Figures
Another abiotic factor that is important in many aquatic 1 and 2. As blood-based gas exchange and skimming gave way
habitats is current drag, which creates a net downstream trans- to tracheal-based gas exchange and three-dimensional flight,
port of stream-dwelling insects during their lifetime. Down- insects that fed on pollen, seeds, and leaves on the water surface
stream drift must be counteracted at some point in the life may have evolved into arboreal plant feeders, thus triggering
cycle by upstream dispersal (otherwise all populations end up the rapid and extensive coevolution of plants and their insect
in the ocean), which modern insects accomplish by flying up- pollinators, seed predators, and herbivores.
stream during the adult stage (Müller 1982). This need not be This is a grandiose evolutionary scenario derived from a fairly
a group-selection argument, since upstream regions can be limited set of data. Many other scenarios have been proposed
richer in oxygen and food, which would allow higher rates of previously (e.g., Kingsolver and Koehl 1985; Kukalova-Peck 1991;
egg and nymph survival. Before the evolution of flight, up- Brodsky 1994; Leech and Cady 1994; Dudley 2000), and it has
stream dispersal was probably accomplished by slow movement been estimated that there have been perhaps 106 words published
along the bottom (where exposure to predators may have been in the scientific literature on the subject of insect wing and flight
high) or along the water surface (where current drag is greatly origins based on only about 102 or 103 data points (Kingsolver
reduced and where predators may have initially been nonex- and Koehl 1994). Thus, our presentation of yet another scenario
istent). Current drag on the surface could have been reduced might be perceived as having limited utility. However, our model
by using the legs to elevate the anterior portion of the body has certain unique features that can potentially make important
above the water (Fig. 1C) while thrust was generated by axial contributions to both the current understanding and the way
bending (Fig. 1B) and flapping of the abdominal gills. Elevation future workers approach the problem. One of these is an em-

This content downloaded from 128.240.233.146 on Sun, 10 Jan 2016 19:12:41 UTC
All use subject to JSTOR Terms and Conditions
 Surface-Skimming Stoneflies and Mayflies 763

phasis on understanding the range of variation in locomotor Cheng L. and H. Hashimoto. 1978. The marine midge Pon-
behavior and mechanics of relatively basal insects, a feature that tomyia (Chironomidae) with a description of females of P.
has been largely absent from previous models of flight evolution. oceana Tokunaga. Syst Entomol 3:189–196.
A second important contribution is that our model does not Dickinson M.H., S. Hannaford, and J. Palka. 1997. The evo-
assume that the evolution of winged locomotion in insects nec- lution of insect wings and their sensory apparatus. Brain
essarily involved a terrestrial stage. A strict equation of wings and Behav Evol 50:13–24.
terrestriality, which is characteristic of all previous models, is Dudley R. 1998. Atmospheric oxygen, giant Paleozoic insects
perhaps best exemplified by the following train of thought from and the evolution of aerial locomotor performance. J Exp
the broad-minded review presented by Kingsolver and Koehl Biol 201:1043–1050.
(1994). They carefully developed the argument that “the current ———. 2000. The Biomechanics of Insect Flight: Form, Func-
evidence appears to support the pleural origin of articulated tion, Evolution. Princeton University Press, Princeton, N.J.
wings on both the abdominal and thoracic segments,” that “initial Ellington C.P. 1984. The aerodynamics of hovering insect flight.
evolution of the protowings occurs in aquatic nymphs,” and that Philos Trans R Soc Lond 305B:1–181.
“the abdominal and thoracic winglets were then subsequently Friedrich M. and D. Tautz. 1995. Ribosomal DNA phylogeny
utilized for some function in the terrestrial environment” (King- of the major extant arthropod classes and the evolution of
solver and Koehl 1994, p. 430). As this thought process dem- myriapods. Nature 376:165–167.
onstrates, even scenarios based on the wings-from-gills model Garcia-Machado E., M. Pempera, N. Dennebouy, M. Oliva-
have been wedded to the assumption that wings are useful only Suarez, J.C. Mounolou, and M. Monnerot. 1999. Mitochon-
in a terrestrial environment. In contrast, our model suggests that drial genes collectively suggest the paraphyly of Crustacea
winged locomotion may have progressed to a flight-capable stage with respect to Insecta. J Mol Evol 49:142–149.
before pterygotes ever became terrestrial or airborne. By pre- Glasheen J. and T. Mcmahon. 1996. Size-dependence of water-
senting a synthetic model that challenges conventional thinking, running ability in basilisk lizards. J Exp Biol 199:2611–2618.
we aim to stimulate new and possibly fertile perspectives. Gould S.J. 1985. Not necessarily a wing: which came first, the
function or the form? Nat Hist 94:12–25.
Graham J.B., R. Dudley, N.M. Aguilar, and C. Gans. 1995.
Acknowledgments
Implications of the late Paleozoic oxygen pulse for physiology
and evolution. Nature 375:117–120.
E. Tsyrlin generously provided logistical assistance, natural his-
Grodnitsky D.L. 1999. Form and Function of Insect Wings:
tory guidance, and species identification during our fieldwork
The Evolution of Biological Structures. Johns Hopkins Uni-
in Australia. J. Earle identified stonefly species collected in
versity Press, Baltimore.
Pennsylvania. This research was supported by National Science
Hennig W. 1981. Insect Phylogeny. Wiley, Chichester.
Foundation grant IBN-9722196.
Hinton H.E. 1981. The Biology of Insect Eggs. Pergamon,
Oxford.
Literature Cited Illies J. 1965. Phylogeny and zoogeography of the Plecoptera.
Annu Rev Entomol 10:117–140.
Aguinaldo A.M. and J.A. Lake. 1998. Evolution of multicellular Jockusch E.L. and L.M. Nagy. 1997. Insect evolution: how did
animals. Am Zool 38:878–887. insect wings originate? Curr Biol 7:R358–R361.
Averof M. and S. Cohen. 1997. Evolutionary origin of insect Kingsolver J.G. and M.A.R. Koehl. 1985. Aerodynamics, ther-
wings from ancestral gills. Nature 385:627–630. moregulation, and the evolution of insect wings: differential
Bock W. 1986. The arboreal origin of avian flight. Pp. 57–81 scaling and evolutionary change. Evolution 39:488–504.
in K. Padian, ed. The Origin of Birds and the Evolution of ———. 1994. Selective factors in the evolution of insect wings.
Flight. California Academy of Science, San Francisco. Annu Rev Entomol 39:425–451.
Boore J.L., D.V. Lavrov, and W.M. Brown. 1998. Gene trans- Kramer M.G. and J.H. Marden. 1997. Almost airborne. Nature
location links insects and crustaceans. Nature 392:667–668. 385:403–404.
Boudreaux H.B. 1979. Arthropod Phylogeny, with Special Ref- Kristensen N.P. 1991. Phylogeny of extant hexapods. Pp.
erence to Insects. Wiley, New York. 125–140 in I.D. Naumann, ed. The Insects of Australia. 2d
Brodsky A.K. 1994. The Evolution of Insect Flight. Oxford ed. Melbourne University Press, Melbourne.
University Press, Oxford. Kukalova-Peck J. 1978. Origin and evolution of insect wings
Burgers P. and L.M. Chiappe. 1999. The wing of Archaeopteryx and their relation to metamorphosis, as documented by the
as a primary thrust generator. Nature 399:60–62. fossil record. J Morphol 156:53–125.
Carroll S.B., S.D. Weatherbee, and J.A. Langeland. 1995. Home- ———. 1983. Origin of the insect wing and wing articulation
otic genes and the regulation and evolution of insect wing from the arthropodan leg. Can J Zool 61:2327–2345.
number. Nature 375:58–61. ———. 1987. New Carboniferous Diplura, Monura, and Thy-

This content downloaded from 128.240.233.146 on Sun, 10 Jan 2016 19:12:41 UTC
All use subject to JSTOR Terms and Conditions
764 J. H. Marden, B. C. O’Donnell, M. A. Thomas, and J. Y. Bye

sanura, the hexapod ground-plan, and the role of thoracic and a new hypothesis for the origin of hexapods. Mol Biol
side lobes in the origin of wings (Insecta). Can J Zool 65: Evol 14:902–913.
2327–2345. Ruffieux L., J. Elouard, and M. Sartori. 1998. Flightlessness in
———. 1991. Fossil history and the evolution of hexapod mayflies and its relevance to hypotheses on the origin of
structures. Pp. 141–179 in I.D. Naumann, ed. The Insects insect flight. Proc R Soc 265B:2135–2140.
of Australia. 2d ed. Melbourne University Press, Melbourne. Samways M.J. 1996. Skimming and insect evolution. Trends
Labandeira C.C. 1994. A compendium of fossil insect families. Ecol Evol 11:471.
Milw Public Mus Contrib Biol Geol 88:1–83. Schmidt-Nielsen K. 1990. Animal Physiology: Adaptation and
Leech R. and A. Cady. 1994. Function shift and the origin of Environment. Cambridge University Press, Cambridge.
insect flight. Aust Biol 7:160–168. Sereno P.C. 1999. The evolution of dinosaurs. Science 284:
Lehmann F.-O. and M.H. Dickinson. 1997. The changes in 2137–2147.
power requirements and muscle efficiency during elevated Shipman P. 1998. Taking wing: Archaeopteryx and the evolution
force production in the fruit fly, Drosophila melanogaster. J of bird flight. Simon & Schuster, New York.
Exp Biol 200:1133–1143. Strausfeld N.J. 1998. Crustacean-insect relationships: the use of
Livezy B.C. and P.S. Humphrey. 1985. Flightlessness in steamer- brain characters to derive phylogeny amongst segmented in-
ducks (Anatidae: Tachyeres): its morphological bases and vertebrates. Brain Behav Evol 52:186–206.
probable evolution. Evolution 40:540–558. Strausfeld N.J., L. Hansen, L. Yongsheng, R.S. Gomez, and K.
Marden J.H. and M.G. Kramer. 1994. Surface-skimming stone- Ito. 1998. Evolution, discovery, and interpretations of ar-
thropod mushroom bodies. Learn Mem 5:11–37.
flies: a possible intermediate stage in insect flight evolution.
Swofford D.L. 1999. PAUP*: Phylogenetic Analysis Using Par-
Science 266:427–430.
simony (*and Other Methods), Version 4. Sinauer, Sunder-
———. 1995. Locomotor performance of insects with rudi-
land, Mass.
mentary wings. Nature 377:332–334.
Thomas A.L.R. 1996. Skimming and insect evolution: reply.
Matsuda R. 1970. Morphology and evolution of the insect tho-
Trends Ecol Evol 11:471.
rax. Mem Entomol Soc Can 76:1–334.
Thomas A.L.R. and R.A. Norberg. 1996. Skimming the surface:
Müller K. 1982. The colonization cycle of freshwater insects.
the origin of flight in insects? Trends Ecol Evol 11:187–188.
Oecologia 52:202–207.
Thomas M.A., K.A. Walsh, M.R. Wolf, B.A. McPheron, and
Nelson C.H. 1984. Numerical cladistic analysis of phylogenetic
J.H. Marden. 2000. Molecular phylogenetic analysis of evo-
relationships in Plecoptera. Ann Entomol Soc Am 77: lutionary trends in stonefly wing structure and locomotion
466–473. behavior. Proc Natl Acad Sci USA (in press).
Ostrom J. 1979. Bird flight: how did it begin? Am Sci 67:46–56. Toms R.B. 1984. Were the first insects terrestrial or aquatic?
Padian K. and L.M. Chiappe. 1998. The origin and early evo- South Afr J Sci 80:319–323.
lution of birds. Biol Rev Camb Philos Soc 73:1–42. Weis-Fogh T. 1973. Quick estimates of flight fitness in hovering
Pritchard G., M.H. McKee, E.M. Pike, G.J. Scrimgeour, and J. animals, including novel mechanisms for lift production. J
Zloty. 1993. Did the first insects live in water or in air? Biol Exp Biol 59:169–230.
J Linn Soc 49:31–44. Will K.W. 1995. Plecopteran surface-skimming and insect flight
Rayner J.M.V. 1991. Avian flight evolution and the problem of evolution. Science 270:684.
Archaeopteryx. Pp. 183–212 in J.M.V. Rayner and R.J. Woot- Williams T.A. 1999. Morphogenesis and homology in arthro-
ton, eds. Biomechanics in Evolution. Cambridge University pod limbs. Am Zool 39:664–675.
Press, Cambridge. Winnepenninckx B.M.H., Y. Van de Peer, and T. Backeljau.
Regier J.C. and J.W. Schultz. 1997. Molecular phylogeny of the 1998. Metazoan relationships on the basis of 18S rRNA se-
major arthropod groups indicates polyphyly of crustaceans quences: a few years later … . Am Zool 38:888–906.

This content downloaded from 128.240.233.146 on Sun, 10 Jan 2016 19:12:41 UTC
All use subject to JSTOR Terms and Conditions