The histology and ultrastructure of the major components of the female reproductive tract in Drosophila (the ovaries, ovarioles, lateral and common oviducts and the uterus) have been outlined previously 910 and here we provide particulars of the musculature. Throughout the tract, the muscle fibres are striated (see Figs. 3, 4) with a clearly defined sarcomeric structure, typical of visceral muscles in insects 11. There is no evidence to support a claim 9 that the epithelial and peritoneal sheath fibres are 'smooth' (presumably by analogy to vertebrate smooth muscle).

Each ovariole is surrounded by an epithelial sheath (Fig. 3A,B, first described in Drosophila 9, that contains bands of circular muscles that run around the ovarioles. There are no muscle fibres running parallel to the long axis of the ovarioles.

In Drosophila, the ovarioles are held together to form two ovaries by an enveloping peritoneal sheath that is a network of muscle fibres (Fig. 3A,B) forming an open mesh around the ovary. Scanning electron micrographs 10 show the presence of large gaps (evident in Fig. 3A) in the peritoneal sheath below which the epithelial sheaths of the ovarioles can be seen.

EM sections cut longitudinally to myofibres of the peritoneal sheath (Fig. 3C) or of the two neighbouring epithelial sheaths (Fig. 3D) show the rather irregular interdigitation of the thick and thin filaments and the presence of perforated Z-discs. These are both characteristic of muscles that are supercontractile (i.e. they contract to less than 50% of their resting length when the thick filaments go through the Z-discs 11). Supercontractile muscles occur in the insect viscera and in the larval body wall musculature 11. This is the first indication that these peritoneal sheath muscle fibres may be supercontractile, but this is not surprising given the large volume changes that occur during post-eclosional maturation of the ovary and oogenesis.

The muscles in the walls of the common (Fig. 4A–C) and lateral oviducts (not shown) are indistinguishable. Confocal fluorescence microscopy (Fig. 4C) shows that they are striated muscles consisting of an ordered array of circular myofibres forming an almost continuous sheet around their respective lumens. Occasional twists and splits are seen in the pattern, especially around the dark "holes" that may contain the nuclei. Ultrastructurally (Fig. 4A,B) the oviduct myofibrils are significantly thicker in cross section than those in the peritoneal and epithelial sheaths around the ovariole, but are structurally identical (Fig. 3). The perforated Z-discs are very clear (Fig. 4B). Infrequently isolated areas are seen in which thick and thin filaments (myofilaments) are cut in cross-section. This may reflect the observation that not all the myofibrils within a fibre show exactly the same orientation (Fig. 4C). Although myofilaments may occasionally deviate from the fibre's long axis, the overall structure shown by confocal fluorescence microscopy did not reveal any fibres or myofibrils running longitudinally in the wall of the oviduct. The oviduct lumen (Fig. 4A) is lined by an epithelial layer (EL) that shows convoluted intracellular membranous structures and extensive microvilli on the apical surface. It seems likely that these are involved in the transport of ions and various molecules 12 to produce oviduct secretions that facilitate egg movement. The epithelial layer has separated from the muscle layer in this sample, which may indicate that the two layers are not strongly adherent in vivo. The circular myofibrils and muscle layer of the uterus (Fig 4D,E) are much more substantial than those of the oviducts and there is no evidence of longitudinal myofibres. Within a single myofibre, neighbouring myofibrils often appear to be in approximate register (Fig. 4E) and the myofibres seem to have tapered ends. It is not known whether these represent attachment sites through which the muscle forces are applied to the uterus or whether this occurs along the length of the myofibres. Ultrastructural studies show (Fig 4D) the sarcomeric structure of the myofibrils, again with perforated Z-discs. The muscle layer is exterior to the epithelial lining of the uterine lumen. The cells in the uterine epithelium, like those of the oviduct (Fig 4A), show a considerable amount of convoluted membranous structures consistent with secretory activity.

In all the regions of the female reproductive tract examined, we found no evidence for longitudinal muscle fibres, though occasional twisting, oblique or longitudinal myo fibrils were seen in the oviduct. Our confocal and ultrastructural observations show that the fibres are striated and their myofibrils have perforated Z-discs indicating that all the muscles are supercontractile 11.

The female genital tract is bilaterally innervated by two pairs of nerves that branch separately from the abdominal median nerve trunk (AbNvTr) 13. One nerve projects to the junction of the ovaries and lateral oviduct (AbNvOv, Fig. 1), where it branches repeatedly. From here the nerves radiate anteriorly across the surface of the peritoneal sheath 4. Fig. 5 shows the nerve fibres have a wandering appearance as they follow the peritoneal sheath muscle fibres. They terminate on the peritoneal sheath, mostly in the posterior two-thirds of the ovary. Confocal Z-stacks show the nerves remain outside the peritoneal sheath muscle layer; they do not dive into the gaps in the peritoneal sheath or contact the ovariolar epithelial sheath. Varicosities are observed at the ends of the nerves, and frequently also along their branches (Fig. 5B). These were often oval and did not look like the "strings of pearls" typical of type I neuromuscular junctions seen on larval or adult body wall muscles 1415, and no glutamate receptor immunoreactivity was seen. The Shaker gene product (a post-synaptic potassium channel) was not detected on the surface of the peritoneal sheath or the ovarioles (Fig. 5C), although myosin was expressed in both. As the Shaker-GFP signal was driven with a Mhc driver, this suggests that the nerves do not construct Type I boutons. All the nerves on the surface of the peritoneal sheath show GFP fluorescence driven by the dTdc2 driver (Fig. 5D). The dTdc2 gene normally expresses the neuron-specific tyrosine decarboxylase. We did not find any HRP immunoreactive fibres without also detecting dTdc2 driven GFP fluorescence. The nerve endings and varicosities all stain brightly with antisera to the Drosophila vesicular monoamine transporter (DVMAT) with some staining visible between varicosities (Fig. 5E). The AbNvOv also innervates the lateral oviduct and the upper portion of the common oviduct. The lateral oviduct has the same type of innervation as the ovary, with fibres showing a similar pattern of varicosities and large endings with the same cytochemistry.

On the common oviduct, a dual pattern of innervation is present. Wandering fibres (Fig. 6A,B) similar to those on the peritoneal sheath and lateral oviduct are present and they co-localise with the dTdc2 gene marker (Fig. 6C), but show no glutamate receptor staining (Fig. 6D). The second type of fibre often runs along the common oviduct (Fig. 6A), and has branches running circularly parallel to the myofibres (Fig. 6A, marked c), (though it is possible that some of the finer parts result from immunoreaction with the muscle cell surface). These neurons do not show dTdc2 driven GFP fluorescence or anti-DVMAT immunostaining, or have the oval varicosities, but rather are associated with round blebs (Fig. 6B). Many of these blebs are correlated with GFP-label from the Shaker potassium channel construct, though this fluorescence is in a focal plane closer to the muscle layer. These blebs are also associated with glutamate receptor immunoreactivity (Fig. 6D).

The extrinsic and intrinsic muscles of the uterus are innervated by a second pair of nerves, branching from the abdominal median nerve trunk. This pair of nerves (AbNvUt, Fig. 1) innervates both the extrinsic muscles and the circular myofibres (data not shown). These nerves have small boutons, closely apposed to larger, round blebs of Shaker-fluorescence, suggesting type I-like terminals. One branch projects to the inner layers of the uterus and shows our dTdc2 marker.

Ovariograms were made by digital video microscopy of the isolated reproductive tract placed on a microscope slide. Data were analysed by overlaying lines on the computer image, and the movement of the light/dark interface along the line was determined (Fig. 2). Fig. 7 shows the analysis of contractions of 4 ovarioles, the peritoneal sheath (PS), oviduct and a spermatheca. The ovariograms consistently show that the isolated genital tract is spontaneously active. The ovarioles, peritoneal sheath, oviduct and spermathecae contract rhythmically and for the most part independently. Oviduct contractions are often associated with those of the peritoneal sheath; they last 1–2 seconds and have a rapid rise and fall. However, oviduct movements, rather than contractions, are also correlated with that of the spermatheca, owing to their strong mechanical coupling. These are much more gradual and last for 10–15 seconds.

Each ovariole moves independently within the peritoneal sheath, with the mean rate being 12.5 ± 6.4 contractions/minute (mean ± 95% confidence interval; mode 10 contractions/minute; 17 ovarioles measured in 6 preparations). There is a very diverse pattern of contractions: some move almost continuously, with up to 53 contractions/minute (Fig. 7A pink trace), others with regular bouts of activity and others remaining largely quiescent (light blue trace, 5 of the 17 ovarioles showed less than 2 contractions/minute). As shown by the traces in Fig. 7(A,B), different ovarioles with a single ovary behave quite differently

The largest recorded movements are those of the spermathecae. These are on the termini of a long thin duct, which contracts or rotates the spermatheca, so that its movements are always very large. In 7-day-old flies these show regular contractions (1.5 ± 0.29/minute, N = 6) with individual contractions lasting 5–6s.

Contractions of the peritoneal sheath are largest at the base of the ovary and so most easily recorded here. In 7-day-old flies the mean frequency of contractions was 5.7 ± 1.6/minute (in preparations in which the spermatheca and seminal receptacle and uterus had been removed to avoid mechanical coupling). In a few flies, we observed eggs leaving the ovary; the progress of the egg was correlated with bursts of contraction. The signal/noise ratio for the peritoneal sheath contractions is quite low in 7-day-old flies, when measuring the position of the light/dark boundary. Using the mean square intensity method gives good clear recordings of the peritoneal sheath waveform (Fig. 2), while in 3-day-old flies, the amplitude of contractions is bigger than in 7-day-old flies because the eggs are less developed and so the sheath is less taut. Consequently the signal/noise ratio is better and the waves are clear to see (Fig. 8). We isolated ovaries from 3-day-old flies and used their ovariograms (Fig. 8) to investigate the regularity of the peritoneal sheath contractions. In at least half the preparations, wavepaths across the peritoneal sheath vary, with some waves having a single contraction peak and others multiple peaks (Fig. 8 Aii). By overlaying each wave, aligning the peaks for the tip, we found that some waves occurred at the tip before the base, while other moved from base to tip, and others had a complex biphasic wavepath (Fig. 8iii). In other preparations, we observed a consistent pattern, for example in Fig. 8B, where the wave of contraction consistently moved from the tip to the base.

Individual ovarioles teased from ovaries still contract spontaneously. Each has its own pattern of bursts, with waves moving both backwards or forwards along the ovariolar epithelial sheath (Fig. 9). This is unlike the observations reported from Drosophila hydei where waves always moved from the base (pedicel) to the tip of the ovariole 16. The mean contraction frequency of isolated ovarioles, 2.9 ± 1.0 contractions/minute is about one third of their activity within the intact ovary.

The Fast Fourier Transform (FFT, Fig. 7B) reveals that the spermathecae have the greatest power density at 2.8 contractions/minute, owing to their freedom to move with large amplitude as their ducts contract and rotate. The spermatheca FFT has prominent sidebands, as would be expected from their tendency to remain contracted for 5–6s. The oviduct (and the peritoneal sheath) share many peaks with the spermatheca, as they are tightly mechanically coupled in this isolated preparation. The peritoneal sheath contractions show a series of small peaks in the spectrum from 10 to 40 contractions/minute, reflecting endogenous peritoneal sheath contractions, some shared with the oviduct. The ovarioles have a different peak power density, in the range 3 to 16 contractions/minute.

We have begun our pharmacological analysis by examining the response of the peritoneal sheath to octopamine and tyramine because the contractions of the peritoneal sheath are an essential first step in the movement of the egg along the reproductive tract. We used the ovary and oviduct preparation to avoid contamination of the traces by spermathecal or uterine contractions. The mean duration of the samples was 145 ± 8 s.

Our ovariograms from 7-day-old flies show that the amplitude of the peritoneal sheath contractions at the base of the ovary increases with 100 nM octopamine (81% increase, P < 0.02, N = 6) (Fig. 10). In contrast, tyramine application at doses up to 1 μM has no significant effect on the amplitude (P = 0.77, N = 12). After application of the neuromodulator, preparations were washed and the contraction amplitudes measured to be 83% of the control value (P = 0.3, N = 31).

The mean frequency of peritoneal sheath contractions from 7-day-old flies (5.7 ± 1.6 contractions/min) did not change significantly following application of octopamine or tyramine, either by comparison of all preparations or by paired comparison of individual preparations (Fig. 11A). Contractions typically occur in bouts, which occur every 0.3–2.5 minutes. We used a minimum interburst interval of 0.2 minutes and determined the mean frequency of bouts to be 1.1 ± 0.28 bouts/minute. The occurrence of bouts was also not affected by either of these drugs (Fig. 11B).

The female reproductive tract is a muscular system, in which contractions are linked to movement of the eggs down the tract from ovarioles via the oviducts to the uterus and finally oviposition of the fertilised eggs. We have described circular fibres in the muscle layers of tubular tissues of the Drosophila reproductive tract (ovarioles, oviduct and uterus). In the walls of the larval and adult gut and in some other insect duct systems, the layers of circular muscles are frequently opposed by longitudinal muscles 11. Longitudinal fibres were apparently detected during development of the Drosophila ovary 9, and have been found, along with circular myofibres, in the ovaries of a number of Lepidoptera including the silk moth Hyalophora cecropia 17, the flour moth Ephestia kühniella 18, the sugar cane borer Diatraea saccharalis 19 and the butterfly Calpodes ethlius 20. However, we have found no evidence for the presence of longitudinal fibres in the female Drosophila genital tract using electron microscopy, immunocytochemistry or directed GFP expression. This is consistent with the movements measured in our ovariograms, where peristaltic waves, but not shortening, were observed. The Drosophila ovary is covered by a peritoneal sheath, which as noted by others 910 contains a network of muscle fibres. Similar sheaths have been reported from other Diptera 21 and may perform a similar function to that of the longitudinal fibres described in other insects; the Lepidopteran ovaries lack peritoneal sheaths. The neural co-ordination of muscle contractions in the Drosophila peritoneal sheath and various tubular muscles are likely to be important for egg development and oviposition. These contractions may also be important for moving the haemolymph around the reproductive tract, especially within the ovary during the energetically and nutritionally demanding process of oogenesis.

Our anti-HRP immunostaining showed that the nerves to the ovary run along the peritoneal sheath network, outside the muscle layer rather than among its myofibrils. We have detected no direct innervation of the myofibres of the ovariolar epithelial sheath. Thus it seems likely that the contractions of the ovariole sheath are controlled by a (local) hormonal mechanism. This would fit with the persistence of spontaneous myogenic contractile activity in the isolated ovariole.

The peritoneal sheath and lateral oviduct have a common pattern of innervation, in which the nerve fibres make oval varicosities, both along their length and at their endings. This is much closer to the type II neuromuscular junction boutons described in adult thoracic muscles 15 than to the type I. We propose that they are modulatory terminals: a suggestion supported by the lack of the Shaker channels, which are usually seen opposite type I terminals 22, as is glutamate receptor immunostaining. The lack of Shaker and glutamate fluorescence was not due to poor staining as the common oviduct and uterus of the same specimens consistently showed them at similar confocal settings. We have direct evidence that these nerves are all aminergic using two markers for tyramine and octopamine synthesising neurons. First, we used a brighter GFP to extend previous data 4, showing the presence of dTdc2 GFP marker on the peritoneal sheath 4 by demonstrating that this marker highlights not just some but all the nerves running over the surface of the peritoneal sheath. As tyrosine decarboxylase (TDC) is essential for the production of both tyramine and octopamine the implication is that these are all tyramine and/or octopamine releasing neurons. Secondly, we find that all the terminals are immunostained with the Drosophila vesicular monoamine transporter (DVMAT) antiserum 23. There is a difference in intracellular location of dTdc2 GFP and DVMAT staining: the dTdc2 GFP co-localises with HRP marker along the length of the neurons, while the DVMAT is only seen at and between varicosities. The ovarial innervation was detected with an antiserum to tyramine β-hydroxylase, the enzyme that converts tyramine to octopamine 24. An antiserum to the OAMB octopamine receptor shows staining in the lateral and common oviducts, in the peritoneal sheath and possibly in the ovarioles 6. Taken together these observations further support the contention that these neurons release octopamine and or tyramine at the peritoneal sheath and our ovariograms confirm that octopamine increases peritoneal sheath contractions. Although the fibres seen by these approaches are the same as we find by anti-HRP staining, it remains possible that not all the neurons use tyramine/octopamine, and that a proportion of the neurons use other transmitters or use a co-transmitter (e.g. a peptide) to modulate the activity of the ovary. Anatomically similar endings have been described on the hindgut and shown to be proctolin-immunoreactive 25.

The common oviduct and uterus show a second pattern of innervation, where nerves, ending as round boutons, on the muscle fibre layers were common. Nerve and muscle are arranged systematically, with nerve branches running parallel to the circular muscle fibres. Here we found glutamate receptors (GluRA; though other subtypes may be present). Our detection of these receptors and of the post-synaptic Shaker channel under the blebs, both typical of adult type I neuromuscular junctions 15, coupled with synaptotagmin immunostaining of the fine endings in the muscle layer 26, suggests conventional neuromuscular junctions rather than those associated with neurohormonal release. In the oviduct, we also observed modulatory endings very similar to those found on the peritoneal sheath, in agreement with the dTdc2 or tyramine β-hydroxylase staining 424. Modulatory endings in the luminal layers of the uterine muscle have not been reported previously.

Observations of movement in isolated ovaries has been reported previously using casual observations or videotape 1621. Our technique to quantify the movement of the Drosophila reproductive tract shows persistent, steady recordings in HL-3 saline. Discrimination of movement of different parts of the reproductive tract shows that the ovarioles, the peritoneal sheath of the ovary, oviduct and spermatheca all generate independent contractions. In each organ, waves of contraction come in bursts, separated by longer quiescent periods. These have all been recorded in isolation from the CNS and so are myogenic in origin. This is confirmed by the persistence of ovariolar rhythms in the isolated ovariole, and by the continued contraction of the peritoneal sheath in the isolated ovary. Myogenic rhythms have been recorded from a range of insect visceral tissues, including the gut and reproductive organs. In the gut, different regions display independent rhythms, where bath-applied amines and peptides modulate large changes in the amplitude and frequency of contractions 2728.

Our data indicate that octopamine significantly increases the amplitude of the peritoneal sheath contractions. In locusts (Locusta migratoria), octopamine reduces the spontaneous contractions of the oviduct 2930. It diminishes both the frequency of the myogenic rhythm and basal tone. In addition, octopamine reduces the proctolin-induced contractions of the oviduct, with 50% block between 10^-6 and 10^-7 M. Similar data come from a dipteran, the stable fly (Stomoxys), in which octopamine also causes a reduction in the spontaneous and proctolin-induced contractions of the oviduct 31.

Our data have led us to a new model in which tissue-specific differences are important. Octopamine increases the strength of contractions of the peritoneal sheath, which will facilitate ovulation of the egg. At the same time, we expect octopamine to relax the oviduct, so that the egg can move into it more easily. Thus, one neuromodulator acts on two neighbouring zones with opposing effects to enhance reproductive success. This hypothesis is in accord with the locust oviduct data, and also with Stomoxys, where 10–100 nM octopamine increases the amplitude of ovarian contractions but relaxes the oviduct 2131.

This model resolves a paradox. Previous authors found it hard to reconcile observations that flies lacking octopamine (owing to the M18 null mutation in the tyramine β-hydroxylase gene) or with a null mutation in the Oamb type of octopamine receptor lay fewer eggs than wild type 624 with the fact that octopamine reduces reproductive tract contractions. Our observations provide a simple explanation for the loss of egg-laying in octopamine-free flies: we suggest that octopamine is likely to be released during egg-laying behaviour to enhance the strength of ovarian contractions and to relax the oviduct and so speed the egg along the reproductive tract. Octopamine may also affect the spermatheca; in locusts the spermathecal contractions were increased by octopamine 32, and the oviduct relaxed 29.

Finally, we showed that tyramine had no effect on the amplitude of the contractions. This is different from the response of the locust oviduct, where tyramine had a very similar inhibitory effect to octopamine on the contraction frequency and basal tone 33. In locust, the effect of octopamine was mediated by cAMP, but the tyramine response did not depend on a cAMP pathway except at very high (10^-4 M) concentrations. This suggests a difference in receptor activation.

In flies, several "octopamine" receptors have been reported. One sensitive to both tyramine and octopamine (TyrR = hono) 3435 is unlikely to be the receptor on the peritoneal sheath, which was only sensitive to octopamine. At least four receptors (Oamb and the DmOctβ receptors 363738) are two orders of magnitude more sensitive to octopamine than tyramine, in accord with our pharmacology, and so may be present on the peritoneal sheath. Although Oamb has been localised on the oviduct 6, the receptor on the peritoneal sheath awaits identification.

Flies were maintained at 25°C on standard yeast-sugar-agar medium. Wild type was Canton-S. Some observations used flies from the Wee-P26 stock 39, which carries a green fluorescent protein (GFP) coding sequence inserted into the myosin heavy chain gene, Mhc. We used a stock in which the CD8-GFP-Shaker protein is driven by a Mhc gene promoter 22. This expresses Shaker-GFP in the post-synaptic densities at the type I neuromuscular junctions of the nerves ending on the muscles. To visualise the synapses of neurons expressing tyrosine decarboxylase 2 (TDC2), we used the dTdc2-GAL4 construct 4 to drive expression of UAS-n-syb-spH, an enhanced GFP fused to neuronal synaptobrevin 40.

Reproductive tracts from 7-day old females were fixed at room temperature for 30 min in 4% paraformaldehyde in phosphate buffered saline (PBS) or for 3 minutes in Bouin's solution (Sigma-Aldrich, UK), washed twice with PBS and incubated in two changes of 0.5% Triton-X100 in PBS for 30 minutes. After further PBS washes, specimens were treated with antibodies or phalloidin-TRITC. Some treatments were applied sequentially to provide a double label. Dilutions and washes were made in PBS.

Actin was labelled by incubation in 50 nM TRITC-labelled phalloidin (Sigma-Aldrich) for 2 h at room temperature. Myosin was labelled by incubation in rabbit anti-myosin heavy chain antibody (1:1000) 41 overnight at 4°C, followed, after washing, by a further 3 h incubation at room temperature in FITC-conjugated goat anti-rabbit IgG antibody (1:250) (Sigma-Aldrich). Vesicular monoamine transporter was labelled by incubation in rabbit anti-Drosophila vesicular monoamine transporter antibody (anti-DVMAT-A, 1:500; 23) overnight at 4°C followed by washing and incubation in FITC-conjugated goat-anti rabbit IgG antibody as described above. Glutamate receptors were labelled as described 42 by incubation with mouse anti-glutamate receptor (dGluRIIA, 1:5) 43, overnight at 4°C, followed, after washing, by a further 3 h incubation at room temperature in FITC-conjugated goat anti-mouse IgG antibody (1:250) (Sigma-Aldrich). Nerve fibres were labelled by incubation in Cy3-conjugated anti-horseradish peroxidase antibody (anti-HRP, 1:1000, Jackson ImmunoResearch, West Grove, PA, USA) for 2 h at room temperature.

After labelling, all specimens were washed in PBS and mounted in Vectashield (Vector Laboratories, Burlingame, CA, USA). Confocal images were obtained with a Zeiss LSM510 META mounted on a Zeiss Axioplan 2 M microscope.

Samples of female reproductive tract were processed for electron microscopy as described for flight muscle 44.

Contractions of the isolated reproductive tract were recorded using video-microscopy followed by computer analysis of the movements from frame to frame. This approach produces an ovariogram in which the contractions at several sites on the same preparation can be recorded and compared.

Flies were transferred to vials within 24 hours of hatching and used on day 3 or day 7. Each vial contained both males and females, so our data are assumed to be from fertilised females. Complete female genital tracts were dissected into cold calcium-free HL-3 medium 45. In most experiments, the spermathecae and seminal receptacle were removed to avoid the large mechanical coupling between spermatheca and oviduct movements (Figs. 10, 11). In some experiments oviducts were also removed to facilitate ovariogram recordings from isolated ovaries (Fig. 8).

Preparations were mounted in fresh HL-3 saline (Ca concentration 1 mM) at 22°C on cavity microscope slides for 20 minutes before observations began. Data were obtained using a Nikon Labophot microscope with a 4× objective through a 5× photographic eyepiece with a Panasonic WV-CL10 camera. The video recordings were captured directly to a 1 MHz PC equipped with a Hauppauge Win-TV capture card (44805), using the VirtualDub 1.5.3 software package and the Huffyuv lossless compression algorithm (Huffyuv Codec) 46. Resolution was set at 640 × 480 pixels, capturing at 2 or 15 frames/second, average recording time 145 ± 8 s.

Video was analysed using a custom program (avi_line, 47). In this, the mouse was used to overlay lines on the video frames, so that each line crossed the light/dark boundary between the preparation and the background. For each frame, the distance of the light/dark interface from the start of the line was determined, along with the mean squared difference in intensity between successive frames of the pixels along the line (Fig. 2). This records the displacement in the plane of focus, but any movement in the vertical direction is not measured. Data were saved in a Microsoft Excel format and mean peak height and intervals between peaks calculated. The Fast Fourier Transforms (FFT) were calculated by exporting the position of the light/dark interface into the fft program from Physionet 48.

In a few experiments, ovarioles from a single ovary were teased apart and video recorded as above but using a 10× phase contrast objective (Fig. 9).

After the initial 20 minute acclimatisation, the activity of the isolated reproductive tract was recorded for 2–3 minutes. The solution was then changed for one containing octopamine or tyramine and a further 2–3 minutes recorded. The control HL-3 solution was then applied for another 2–3 minutes recording. Because of disk space limitations, we were unable to record for longer, and this may affect the accuracy of our measurements of the frequency of contraction. Amines and HL-3 saline components were from Sigma-Aldrich, UK.

For each preparation, the mean amplitude and interval between contractions were calculated. Differences between these means were tested in Minitab, using Mann-Whitney and Student's t-tests. All data are reported as mean ± 95% confidence interval.