Wing movements during lift-off and flight

Males and females of Thraulodes cochunaensis (Fig. 1, video S1) and T. consortis (Leptophlebiidae), Leptohyphes eximius (Leptohyphidae) (video S2) and Callibaetis guttatus (Baetidae) were high-speed recorded at lift-off and initial flight. We observed the same type of flight also in other families, though without recording them, e.g. in Siphlonurus lacustris (Siphlonuridae), Epeorus assimilis (Heptageniidae), Ephemera danica (Ephemeridae), Serratella ignita (Ephemerellidae), and Caenis gonseri (Caenidae). All recorded specimens essentially have identical patterns of wing cycles. In all four species, the hind wings are highly reduced and hardly visible at all, so we only consider the movement of the forewings in the following description of T. cochunaensis (Fig. 1): At rest, all wings are folded upwards and held vertically above the abdomen with the dorsal sides of the forewings touching each other. The anterior margin of the wing is thereby obliquely directed in a 45° angle in such a manner that the wing tip marks the highest and most posterior point of the forewing (Fig. 1A). The entire wing cycle resembles the typical movement of a rowing paddle, albeit in a vertical rather than horizontal movement. The lift-off is initiated by a firm push-off with all legs, which is accompanied by the first partial downstroke of the wings (Fig. 1B–D). This initial downstroke is not complete in order to prevent the wings from touching the ground. The jump lifts the entire animal high enough to prevent the wings from touching the ground during subsequent wing cycles, when the downstrokes end in the wings held vertically to the longitudinal body axis (Fig. 1R, Y). During two thirds of the downstroke, fore- and hind margins approximately stay at same heights (Fig. 1I,J,P,Q), so the plane of each forewing remains straight to produce maximum uplift. In the last third of the downstroke, the anterior wing margin is put forward and angled compared to the posterior flexible half of the wing, which shows some inertia trailing behind (Fig. 1Q,R,X). To overcome air resistance, the wings during upstroke likewise rotate obliquely along their longitudinal axis when lifted, with the costal margin leading forward and upward, so the wing is held in a vertical position (Fig. 1L, S, AA). Again, the posterior part of the wing slightly cambers and is trailing behind. The anterior area of the wing between veins costa and radius anterior (R1) always remains straight throughout the entire wing cycle, followed by the flexible posterior part of the wing, resulting in a cambering movement throughout the whole wing cycle (see video S1). At no times we could observe any breaking or bending of the wings in the region where the bullae are located.

Fig. 1

Thraulodes cochunaensis (Leptophlebiidae), still photographies from high-speed video S1 at 1677 frames per second showing first six wing cycles at lift-off. A Resting position, B–D initial jump-off and first downstroke, E–G first upstroke, H–K second downstroke, L–N third upstroke, O–R fourth downstroke, S–U fifth upstroke, V–Y fifth downstroke, Z–BB sixth upstroke, CC–DD sixth downstroke

Different flight modes in swarming mayflies

In mayflies, the nuptial flight is performed by swarming male specimens, which aggregate in swarms above or near the water body. We video recorded the swarming in Leptohyphes eximius (Leptohyphidae) (video S3), Miroculis fittkaui (Leptophlebiidae) (video S4), and Lachlania sp. (Oligoneuriidae) (video S5). The males of L. eximius fly in a vertically orientated pattern, which includes a slower ascent of variable length and a rapid descent in one go. Males of M. fittkaui hover more or less stationary, occasionally shooting up very rapidly. In Lachlania sp., the males patrol the stream in an irregular horizontal pattern.

Morphology of wing veins and bullae

Mayfly wings are typically highly corrugated with alternating positive and negative longitudinal veins, which lie either above or below the wing plane, as exemplified here by Ephemera danica (Figs. 2A and 3A). When longitudinal veins fork, they retain their spatial position, and an additional intercalary vein appears in between the forked vein to keep the general alternating corrugation. Also, the first three main longitudinal veins (C, Sc, R1) are the most rigid ones, while the following ones tend to get weaker from R2 to the anal veins (Figs. 2B and 3B). Moreover, in some of the negative longitudinal veins (Sc, R2, R4 + 5, MP1) at approximately half-length, there are bullae present as unsclerotized blisters in both subimagines and imagines (Figs. 2 and 3, see also Table 1 for distribution across Ephemeroptera).

Fig. 2

Ephemera danica (Ephemeridae), forewing and bullae of subimago in ventral view after critical point drying under light microscopy. A Total view, B bullae in detailed view, C bulla sc, D bulla r2, E bulla r4 + 5, F bulla mp1. Abbreviations of veins (positive veins labeled in black, negative in white): c: costa, sc: subcosta, r: radius, ma: media anterior, mp: media posterior, C–F in same magnification

Fig. 3

Ephemera danica (Ephemeridae), forewing and bullae of imago in ventral view after critical point drying under light microscopy. A Total view, B bullae in detailed view, C bulla sc, D bulla r2, E bulla r4 + 5, F bulla mp1. Abbreviations of veins (positive veins labeled in black, negative in white): c: costa, sc: subcosta, r: radius, ma: media anterior, mp: media posterior. C–F in same magnification

Table 1 Numbers and distribution of bullae throughout different taxa of extant and fossil mayflies

The membranous nature of the bullae becomes even more evident after critical point drying of the wing (Figs. 2C–F and 3C–F). These desclerotized, membranous portions of the veins appear also inflated under SEM, especially in ventral view (Fig. 4A–H). In dorsal view, bullae in the SEM are much more inconspicuous (Fig. 4I–L). Bullae in principle have a similar structure in different species (e.g. Siphlonurus croaticus, Fig. 4M–P).

Fig. 4

Bullae sc, r2, r4 + 5, and mp1 under SEM, A–D Ephemera danica, subimago, ventral view, E–H Ephemera danica, imago, ventral view, I–L Ephemera danica, imago, dorsal view, M–P Siphlonurus croaticus, imago, ventral view. A–L and M–P in same magnification

In confocal laser microscopy, the degree of resilin enrichment in the cuticle of Siphlonurus lacustris can be visualized (Fig. 5A) by the intensity of blue colour. Also in all other investigated species, resilin is distributed all over the wing membrane, but neither particularly enriched in the main longitudinal veins nor in the bullae themselves. Membranous and sclerotized properties of the wing veins can also be visualized by different colours, where membranous areas are shown in green colour and sclerotized areas in red colour. It becomes obvious that the sclerotized exocuticular layer is entirely missing in the bullae ventrally (Fig. 5B–F).

Fig. 5

Siphlonurus lacustris (Siphlonuridae), imago, ventral view under confocal microscopy. A Forewing, presence of resilin, B forewing, distribution of sclerotized and membranous areas, C bulla sc, D bulla r2, E bulla r4 + 5, F bulla mp1. Red colour indicates sclerotized cuticle, green colour indicates membranous cuticle, blue colour indicates presence of resilin

µCT scans of the subcostal vein in E. danica reveal that its ventral layer is much thicker than its dorsal counterpart in both subimago (Fig. 6B–D) and imago. The same applies to all other negative longitudinal veins. In contrast, in positive veins it is always the dorsal layer, which is thickened. The only exception is the costal vein, which is in the leading edge of the wing and therefore equally thickened throughout (Fig. 6B–D).

Fig. 6

Ephemera danica (Ephemeridae), subimago, region of bulla sc shown as volume rendering based on µCT data. A ventral view, dashed lines indicate different levels of cross cuts in B–D, E ventral view with longitudinal veins, crossveins and bulla sc indicated. Abbreviations: c: costal vein, sc: subcostal vein, cv1-3: crossveins in costal field, cv4-5: crossveins in subcostal field, bu: bulla, cutSI: subimaginal layer of cuticle, cutI: imaginal layer of cuticle, trab: trabeculae. Without scales

Additionally, in the subimago, the imaginal wing is already preformed and visible within the surrounding subimaginal wing, showing already the same pattern of sclerotization of cuticle. At this developmental stage, between the upper and lower cuticles of crossveins there are multiple dorsoventral, column- or strut-like connections present, reminiscent of palisade parenchyma in plant tissue or trabeculae in vertebrate bones (Fig. 6B, D). These trabeculae are missing after the final moulting in the aerodynamic profile of the thinner imaginal wing. The resolution of the µCT is not sufficient to determine the nature of these trabeculae. As a more detailed investigation on the histology of the wing was beyond the scope of this contribution, we will delve into this matter in a forthcoming separate study.

Distribution of bullae in different extant families of mayflies

We studied the distribution of bullae in numerous species of mayflies throughout the order, representing the presently recognized families within Ephemeroptera (Figs. 7 and 8, Table 1).

Fig. 7

Distribution of bullae in forewings of species from different families. A Siphluriscus chinensis (Siphluriscidae), B Metamonius anceps (Nesameletidae), C Siphlonurus croaticus (Siphlonuridae), D Baetis rhodani (Baetidae), E Chromarcys magnifica (Oligoneuriidae: Chromarcyinae), F Oligoneuriella rhenana (Oligoneuriidae: Oligoneuriinae), G Ecdyonurus venosus (Heptageniidae), H Isonychia berneri (Isonychiidae), I Calliarcys humilis (Leptophlebiidae: Calliarcyinae), J Habroleptoides confusa (Leptophlebiidae: Leptophlebiinae), K Miroculis misionensis ♂ (Leptophlebiidae: Atalophlebiinae), L Miroculis misionensis ♀ (Leptophlebiidae: Atalophlebiinae)

Fig. 8

Distribution of bullae in forewings of species from different families. A Potamanthus luteus (Potamanthidae), B Dolania americana ♂ (Behningiidae), C Ephoron virgo ♂ (Polymitarcyidae), D Ephoron virgo ♀ (Polymitarcyidae), E Eurylophella trilineata (Ephemerellidae), F Serratella ignita (Ephemerellidae), G Leptohyphes cornutus (Leptohyphidae), H Teloganodes sp. (Teloganodidae), I Caenis horaria (Caenidae), J Baetisca berneri (Baetiscidae)

Within Siphlonuroidea, in Siphluriscidae (Fig. 7A) there are six bullae present, approximately at half-length in each of the following veins Sc, R1, R2, R4 + 5, iMA, MP1 (in order from anterior to posterior). All of these veins except R1 are negative veins. Siphluriscus chinensis is the only mayfly, in which we found a clear bulla in iMA. In siphlonuroid families with amphinotic distribution, e.g. in Nesameletidae (Fig. 7B), Oniscigastridae, and Ameletopsidae, bullae are present in the same veins except of iMA. This also applies for Rallidentidae and Dipteromimidae, which are endemic to New Zealand and Japan, respectively. In other siphlonuroid families, which are distributed in the northern hemisphere, there are neither in iMA nor in R1 bullae present, e.g. in Siphlonuridae (Fig. 7C), Ameletidae, Metretopodidae, and Acanthametropodidae. In Ametropodidae, there are only three bullae present, namely in Sc, R2, and R4 + 5.

In Baetoidea, in both Siphlaenigmatidae and Baetidae (Fig. 7D), there are bullae present in Sc, R2, and R4 + 5.

In Heptagenioidea, we found four bullae in Coloburiscidae, Isonychiidae (Fig. 7H), Oligoneuriidae: Chromarcyinae (Fig. 7E), and Heptageniidae (Fig. 7G). In the highly modified wings of Oligoneuriidae: Oligoneuriinae (Fig. 7F), there were no bullae found at all.

In Leptophlebioidea, four to three bullae were found in different species in each of the different subfamilies Leptophlebiinae and Atalophlebiinae: While Calliarcys humilis (Fig. 7I) and Paraleptophlebia submarginata (both Leptophlebiinae) both have four bullae, there are only three bullae present in Habroleptoides confusa (Fig. 7J). Likewise, in Thraulodes consortis (Atalophlebiinae), there are four bullae present, while both sexes of Ulmeritus carbonelli have only three bullae. There may be also sexual dimorphism present like in Miroculis misionensis, in which females (Fig. 7L) are equipped with four, males (Fig. 7K) only with three bullae.

In Ephemeroidea, four bullae were found in Potamanthidae (Fig. 8A) and Ephemeridae (Fig. 3A), while in Euthyplociidae only three bullae are present. In Palingeniidae, males have three bullae, while the three bullae in females are more subtle and largely reduced. Sexual dimorphism can also be found in Polymitarcyidae, where males of Ephoron virgo have three bullae (Fig. 8C), while in females no bullae are obvious (Fig. 8D). In other species like Campsurus cotaxe, bullae are absent in both sexes. No bullae were also found in Behningiidae (Fig. 8B).

In Ephemerelloidea, most of the families have only three bullae, like in Leptohyphidae (Fig. 8G), Coryphoridae, Teloganellidae, Teloganodidae (Fig. 8H), Tricorythidae, Machadorythidae, Dicercomyzidae, Ephemerythidae, and Vietnamellidae. In Ephemerellidae, except of the usual three bullae (e.g. Serratella ignita, Fig. 8F) there were four bullae found in Eurylophella trilineata (Fig. 8E). Due to the poor conservation of the only known adult specimen of Melanemerella brasiliana, we were not able to verify the number of bullae in Melanemerellidae. Likewise, we had no access to Austremerella picta nor to any species of Teloganella sp., so we have no information on the bullae in Austremerellidae and Teloganellidae.

In Caenoidea, in both Neoephemeridae and Caenidae (Fig. 8I), there are three bullae present, in some cases only very subtle or even not recognizable.

In Prosopistomatoidea, Baetiscidae (Fig. 8J) have four bullae, while in the highly modified wings of Prosopistomatidae bullae are absent in both sexes.

Presence of bullae in fossil Ephemeroptera and Ephemerida

We checked the wings of different fossil species of extant families for the presence of bullae, e.g. among others Borinquena parva (Leptophlebiidae) in Dominican amber, Siphloplecton sp. (Metretopodidae) in Baltic Amber, and Burmella paucivenosa (Vietnamellidae) in Burmese amber. All of them displayed just the same number and distribution of bullae like their extant relatives. We also documented the presence of bullae in extinct families and subfamilies of Ephemeroptera. Undescribed species of Hexagenitidae and Baetidae: Palaeocloeoninae in Burmese amber (99 Ma) were equipped with bullae in Sc, R2, and R4 + 5 (Fig. 9A–B and C–D), respectively. We here also record for the first time an undescribed species of Australiphemeridae in Lebanese amber (129 Ma) that bears bullae in Sc, R2, R4 + 5, and MP1. Finally, we were also able to verify the presence of bullae in stemgroup Ephemerida. The hind wing of Protereisma insigne (Permoplectoptera: Protereismatidae) from the Lower Permian of Kansas (272 Ma) shows likewise clearly pronounced bullae in Sc, R2, R4 + 5, and MP1 (for details, see Fig. 9G,H, Table 1).

Fig. 9

Bullae in wings of fossil crowngroup and stemgroup mayflies. A, B Undescribed species of Hexagenitidae † (Ephemeroptera), forewing. Burmese amber, Mid-Cretaceous, ca. 99 ma. C, D Undescribed species of Baetidae: Palaeocloeoninae † (Ephemeroptera), forewing. Burmese amber, Mid-Cretaceous, ca. 99 Ma. E, F Undescribed species of Australiphemeridae † (Ephemeroptera), forewing, first record from Lebanese amber, Lower Cretaceous, ca. 129 Ma. G, H holotype of Protereisma insigne † (Ephemerida: Permoplectoptera: Protereismatidae), hind wing, Lower Permian of Kansas, ca. 272 Ma., courtesy of the Yale Peabody Museum, Division of Vertebrate Paleontology, Yale University, Peabody.yale.edu. Photographs by S. Butts, 2023

Subimaginal moulting and wing extraction

To observe the process of wing moulting and the possible role of bullae therein, subimaginal moultings were video recorded in Chiloporter eatoni (Ameletopsidae) (Fig. 10, video S8), Siphlonurus croaticus and Siphlonurus lacustris (Siphlonuridae), Callibaetis guttatus (Fig. 11, video S9) and Baetodes huaico (Baetidae), Hapsiphlebia anastomosis, Nousia bella and Thraulodes consortis (Leptophlebiidae), Ephemera danica (Ephemeridae) (video S10), Lumahyphes guacra (Leptohyphidae) and Caenis gonseri (Caenidae). Additionally, we observed some aberrant modes of subimaginal moulting in Oligoneuriella rhenana and Homoeoneuria sp. (Oligoneuriidae) (Fig. 12A), and Asthenopus curtus and Tortopsis sarae (Polymitarcyidae) (Fig. 12B), or even loss of subimaginal moulting in the females of Ephoron virgo (Fig. 8D) and Palingenia longicauda (Palingeniidae).

Fig. 10

A–F Different moulting stages from subimago to imago in Chiloporter eatoni (Ameletopsidae), still photographs from video S8. Arrow in E shows wing bending at line predetermined by bullae

Fig. 11

A–F Different moulting stages from subimago to imago in Callibaetis guttatus (Baetidae), still photographs from video S9. Arrow in E shows wing bending at line predetermined by bullae

Fig. 12

Modified moulting in species lacking bullae. A Body moult in Homoeoneuria sp. (Oligoneuriidae) with subimaginal wings remaining unshed. B Additional delamination and fragmentation of subimaginal wing cuticle in Tortopsis sarae (Polymitarcyidae). Arrows point to abdominal exuvia and delaminating wing cuticle

In Callibaetis guttatus, the subimago usually emerges from the last nymphal instar in the afternoon between 3.30 and 5 pm local time (ART, GMT-3). It immediately flies to a nearby place and rests until the subimaginal moult, which takes place on the same night; usually well before dawn between 3 and 4am. The subimago clings to the vegetation in a vertically orientated position with the head up and becomes inactive for some time. Immediately before moulting the animal gets restless performing sudden movements. The actual moulting process (Fig. 11) is initiated by an irregular trembling of the entire animal and by flickering of the wings, which eventually change their posture (see Video S9). From initially being held vertically above the abdomen, the wings are gradually taken down horizontally and then even further backwards and downwards, so that the costal vein is horizontally in line with the abdomen and the wing tip directed posteriorly. In this way, the wings are aligned to the longitudinal body axis to enable an unobstructed, smooth extraction of the imaginal wing. Additionally, the legs are taken backwards to become likewise aligned. At the same time, the thoracic cuticle ruptures dorsomedially along with the epicranial suture of the head, as the emerging imago propels itself forward by peristaltic body movements. As the animal has almost shed its entire body except for the wing tips, tarsi, last abdominal segments, and tail filaments, it bends over its back with the head down. To facilitate the moulting of the wing apex, it then starts lifting its wings. Thereby the entire wings bend along a flexion line predetermined by the position of the bullae. As soon as the wing tip is released, the wing immediately snaps back in its original shape. At the same time, while the subimaginal leg cuticle remains in natural position with the subimaginal claws anchored to the substrate, the imaginal legs are pulled out. Thereby the imaginal legs change from their natural position in a way that all segments of each leg are aligned in a straight line and directed backwards. As soon as the imaginal legs are fully pulled out, they immediately return to their natural position with usual angles between the different leg segments of each leg to become fully functional again. With the imaginal legs in function, the imago bends upwards again and returns to its initial upright position. As a final step, the tail filaments are extracted from the remaining subimaginal cuticle. The same mode of moulting was basically also observed in all other species mentioned above (see also video S10 for Ephemera danica).

In some cases, like in Chiloporter eatoni, under the artificial conditions provided, the animal did not have the choice to moult in a vertical position, but was forced to moult in a horizontal position (Fig. 10, Video S8). Still, the entire moulting process was similar to the one observed in C. guttatus except that the animal did not bend backwards. During moulting, the animal likewise propelled forward, thereby pulling out imaginal wings and legs of the subimaginal cuticle. However, the extraction of the imaginal wing took much more time and appeared to require more effort than moulting in the upright position.

In any case, the bullae determine the position where the wings break in the critical moment of the extraction of the imaginal wing from within the subimaginal one. Nevertheless, the imaginal wing still breaks in the same region even when there are no evident bullae present, like in C. gonseri.

Apart from the usual process of moulting, different modes exist in some taxa: In Oligoneuriella rhenana and Homoeoneuria sp. (Oligoneuriidae), only the body sheds its subimaginal cuticle, while the subimaginal wings remain unshed like an envelope around the imaginal ones (Fig. 12A). In Asthenopus curtus and Tortopsis sarae (Polymitarcyidae) (Fig. 12B), additionally to the body moulting, the subimaginal wing cuticle is delaminated, separates and peels off during flight. In some other species like Dolania americana (Fig. 8B), Ephoron virgo (Fig. 8D), or Palingenia longicauda, the females do not moult at all and remain in the subimaginal stage throughout their winged life.

Cardboard paper models

To test the behaviour of the wing, the wing cardboard model was fixed at its base by holding it firmly with one hand. When mechanical pressure was applied to the apical half from dorsally with the other hand, it bent following a line predetermined by the positions of the bullae. Applying approximately the same amount of mechanical pressure from ventrally, the model was not affected, withstood the pressure and would not bend at all. Applying increased pressure at some point resulted in random bending of the wing (video S11).