Studies on gas exchange in the meadow spittlebug, Philaenus spumarius: the metabolic cost of feeding on, and living in, xylem sap

Spittlebugs are the nymph stage of the froghopper (Cercopoidea: Homoptera) and are so called because of the conspicuous mass of wet, spittle-like foam that they secrete around themselves while feeding on the sap of plants. Spittlebugs produce this foam by incorporating bubbles of air into an anal exudate that consists primarily of excreted xylem fluid – a plentiful by-product of their xylem-feeding habits (Wiegert, 1964). This fluid is further modified by the addition of proteins and mucopolysaccharides secreted by the Malpighian tubules, which are believed to stabilise the structure of the foam (Marshall, 1966, 1973; Mello et al., 1987). Bubbles are produced when the spittlebug extends the tip of its abdomen through the surface of the foam and draws atmospheric air into a groove on the ventral surface of its abdomen (Hamilton, 1982). This sternal groove is formed by overlapping pleural prolongations, and contains the nymph's abdominal spiracles (Guilbeau, 1908; Hamilton, 1982). Then, by submerging the tip of the abdomen into the foamy liquid and contracting the abdomen (Fig. 1A,B), bubbles are expelled. By rapidly repeating this action, a spittlebug can completely cover its body in foam within 15–30 min (Hamilton, 1982).

The adaptive function of the spittle mass is still debated. It has been proposed that it could protect the nymph from desiccation and/or predation (del Campo et al., 2011; Whittaker, 1970), provide a microclimate that buffers temperature (Tonelli et al., 2018), prevent growth of fungi (Li et al., 2015), or even act as a neutraliser of toxic ammonia (Rakitov, 2002). But, regardless of the potential adaptive benefits, living submerged in a mixture of liquid and gas is an unusual respiratory environment for any terrestrial insect. However, no study has yet determined precisely how spittlebugs obtain O₂ and excrete CO₂ while submerged in a wet foam (Wiegert, 1964). There are two main gas exchange strategies open to a spittle bug: either they respire the O₂ trapped within the bubbles that comprise the foam, thus making the foam mass analogous to the air bubbles and compressible gas-gills carried by some diving insects (Turner, 1994), or they rely on a respiratory siphon or snorkel to access atmospheric O₂ as required (Hamilton, 1982; Kosztarab, 1979). The primary goal of this study was to determine which of these strategies are used. Are they ‘foam breathers’ or do they snorkel?

There is one life stage of the spittlebug, however, where they must rely on the diffusion of O₂ through the wet foam to breathe. Before their final moult from nymph to adult, many spittlebug species create a single large bubble within the foam mass (Fig. 1C). Within this ‘bubble chamber’, they moult into the adult froghopper, remaining concealed until their cuticle has hardened (Weaver and King, 1954). Unlike the spittlebug nymph, the adult froghopper has neither a flexible abdomen able to telescope out of the spittle nor the sternal groove, and thus it is unable to access atmospheric O₂ directly while within the bubble chamber.

The mode of gas exchange used by spittlebugs is also of interest when considering their energetic demands. Spittlebugs feed by inserting their stylets into a plant's xylem tissue then drawing up the sap. However, the cohesion–tension theory of sap transport assumes that xylem sap can exist under very high tensions (>1 MPa) (Kim, 2013). If this is the case, then spittlebugs must expend a large amount of energy to produce an even greater negative pressure to draw out this fluid (Kim, 2013; Malone et al., 1999). All feeding suction must be generated by the bug's cibarial pump, which is powered by large muscles in their head (Malone et al., 1999). In addition, xylem sap is very poor in nutrients, so spittlebugs must ingest large quantities to meet their metabolic demands, routinely excreting 150–280 times their own body mass in fluid every 24 h (Horsfield, 1978). As the power (P; W) required to pump a fluid is equal to the volumetric flow rate of the fluid through the pump (Q; m³ s⁻¹) and the pressure (p; N m⁻²) it works against (i.e. P=p×Q), xylem-feeding bugs would appear to be uniquely disadvantaged in their choice of food. Thus, it must be energetically very costly for spittlebugs to extract these large volumes of xylem sap, unless they employ some unknown method to reduce the xylem tension (Kim, 2013; Malone et al., 1999; Novotny and Wilson, 1997). Furthermore, if the increased metabolic requirements of feeding dramatically increased O₂ consumption, then using the spittle mass as its primary oxygen source may limit the bug's ability to feed. However, the metabolic rate of a spittlebug feeding on xylem has never been measured, so the metabolic cost of this activity, and its consequences for gas exchange, remain unknown.

This study examined the gas exchange strategies and metabolic rates of both the nymph and adult of the meadow spittlebug Philaenus spumarius (Linnaeus 1758). Fibre optic O₂ optodes were used to determine whether a decrease in P_O2 was detectable within spittle masses containing a nymph, as would be consistent with the spittlebug respiring O₂ from within the trapped air bubbles. This technique was also used to measure the P_O2 inside the bubble chamber of a moulting final instar. Both stop-flow O₂ respirometry and flow-through CO₂ respirometry were used to determine the resting metabolic rate (RMR; V̇_CO2, V̇_O2) and respiratory quotient (RQ) of both the spittlebugs and adult froghoppers. Flow-through CO₂ respirometry was also used to measure the metabolic rate of spittlebugs during locomotion and while producing spittle and feeding on xylem sap. Once the spittlebugs had constructed a spittle mass within the respirometry chamber, their responses to mechanical taps simulating the presence of a potential predator were also measured. Finally, spittlebugs were forcibly submerged in spittle on a microscope slide coated with an O₂-sensitive fluorophore to test their ability to access O₂ within the foam, and visualise how the P_O2 changed within adjacent bubbles in the foam mass. Altogether, these measurements provide a comprehensive assessment of the energetic costs and gas exchange strategies that spittlebugs employ to live within the unusual respiratory environment of a wet foam.

All spittlebug nymphs and adults used in these experiments were collected by hand from around the University of British Columbia Point Grey campus between May and July in 2017 and 2018. Individuals were identified using the key in Hamilton (1982), and the instars were determined by wing pad development using the key in Weaver and King (1954).

P_O2 transects through masses of P. spumaris spittlebug foam were measured at the Point Grey Campus of the University of British Columbia. Spittle masses of varying size and on various plant species were selected haphazardly. The P_O2 within the spittle mass was determined using a needle-mounted O₂ optode (130 μm diameter fibre, NTH-PST7-O2-L2.5-TS-NS40-0.8-NOP, PreSens GmbH, Regensburg, Germany) held in a 3-axis micromanipulator (M3301R, World Precision Instruments, Sarasota, FL, USA), which was mounted on a metal stake. The stake was driven into the ground next to the plant, and a second clamp attached to the stake was used to hold the plant's stem. The tip of the optode was always inserted into the spittle mass parallel to the stem of the plant, to avoid breaking the sensor. The optode was advanced into the spittle mass in 1 mm increments, from the outer edge of the mass to the centre where the nymph(s) resided, while P_O2 measurements were recorded at each position using a portable fibre optic oxygen meter (Microx4, PreSens GmbH).

Spittlebugs collected in the field were quickly returned to the laboratory, where they were placed on potted sheep sorrel plants (Rumex acetosella) growing in a standard potting compost (Indoor Potting Soil, GardenWorks, BC, Canada) in 6 inch diameter by 4.5 inch deep pots. Spittlebugs were observed feeding on sheep sorrel in the field, and would readily feed and secrete foam masses on the potted plants. Once foam began to be produced, a needle-mounted O₂ optode held by a micromanipulator was again inserted into the freshly produced foam, this time held perpendicular to the plant stem or leaf, with the tip positioned as close as possible to the feeding bug. The optode remained in this position for 2 h while P_O2 was logged every 20 s.

Plant stems and leaves that had bubble chambers containing a moulting final instar nymph on them were removed and taken back to the lab where the P_O2 within the bubble chamber was recorded, again using a needle-mounted O₂ optode. Each bubble chamber was inspected to make sure it was intact and completely covered by a layer of foam. The tip of the O₂ optode was then carefully inserted into the bubble so as to not cause it to collapse or burst, and the P_O2 within the bubble chamber was measured every 20 s for 2 h. At the end of the measurement period, the adult insect was removed from within the bubble, and its weight and length were measured. The reported P_O2 values inside the bubble chambers were averaged from the first 15 min of measurements and were compared with the atmospheric O₂ directly outside the bubble chamber.

The rates of O₂ consumption (V̇_O2) and CO₂ production (V̇_CO2) of spittlebug nymphs and adults at rest were determined in the laboratory. All nymphs were individually weighed to 0.01 mg on an electronic balance (XPE205 DeltaRange, Mettler Toledo, Greifensee, Switzerland) before and after respirometry measurements, and all adults were weighed after. Nymphs or adults were placed individually into 2 ml glass vial chambers (11 mm diameter, 35 mm long) which were then sealed and left for 2 h while the P_O2 or CO₂ concentration within the vial was continuously measured. A small piece of nylon mesh was also placed inside the vial to give the insects a surface to grip onto during the trials. During the measurement period, both the CO₂ and O₂ respirometry chambers were inserted into close-fitting holes bored horizontally into an aluminium block (45 mm×45 mm×146 mm). The block was placed on a temperature-controlled Peltier plate (AHP-1200CPV, ThermoElectric Cooling America Corporation, Chicago, IL, USA) and maintained at a constant temperature of 20°C.

A similar flow-through respirometry system to that described above was used to measure V̇_CO2 during active xylem feeding and spittle production, where a clear acrylic chamber was used instead of the glass vials. The chamber was made from two square (50×50 mm) acrylic blocks with circular holes milled into each half (30 mm diameter and 5 mm deep in one half, 30 mm diameter and 6 mm in the other half, giving an approximate volume of 7.76 ml). The two squares interlocked to make a central circular chamber. A groove (3×3×50 mm) was cut across the midline of each acrylic block, bisecting the circular chamber. This groove allowed both blocks to be clamped around the stem of a plant, forming an enclosed chamber. The two halves of the chamber were sealed around a stem of sheep sorrel using 2-part polyvinylsiloxane casting material (President light body dental impression material, Coltène Whaledent, Altstätten, Switzerland). One half of the chamber had a round glass window epoxied into it for observation using a digital camera (PiNoir IR camera Raspberry Pi 3rd Edition Model B, Allied Electronics Inc., London, UK). The other half had two short lengths of 21-gauge stainless steel hypodermic tubing (8 mm long) epoxied into two holes in the top of the chamber, for the air inlet and outlet. Dry, CO₂-free air was metered at a rate of 50 ml min⁻¹ using a 0–100 ml min⁻¹ mass flow controller (Alicat Scientific). Air was humidified before, and dried after, the respirometry chamber using the same Nafion tubes as described above. The dried excurrent air was passed into a LI-820 CO₂ IRGA gas analyser (Licor) and CO₂ concentration (in ppm), along with chamber temperature measured using a 0.010 gauge T-type thermocouple, were sampled at 2 Hz using a Powerlab 8/35 DAQ (ADInstruments) and recorded using LabChart software (v.8.1.5, ADInstruments).

After recording a CO₂ baseline with a sheep sorrel stem mounted in the chamber, a spittlebug was introduced into the chamber through a port (4 mm diameter) in the acrylic block. This opening was then sealed with putty (UHU GmbH & Co. KG, Bühl, Germany). Spittlebug behaviour and movement were annotated to the LabChart recording in real time by a researcher watching the live video of the spittlebug inside the chamber. Once a spittlebug had secreted a volume of froth that completely covered its body, the chamber was sharply tapped several times with a pair of metal forceps to simulate a potential predator. Every individual that made a spittle mass experienced 2–5 separate tapping events, and any changes in their V̇_CO2 following these ‘attacks’ were recorded.

A novel fluorescence approach was used to visualise whether spittle bugs were capable of extracting O₂ from the air bubbles trapped within the foam, as well as whether diffusion of O₂ between adjacent air bubbles could occur. To do this, glass microscope slides (75×25×1 mm) were coated in a thin silicone elastomer film containing an O₂-sensitive fluorophore: platinum (II) meso-tetra(pentafluorphenyl)porphine (PtTFPP; Frontier Scientific, Newark, DE, USA). The PtTFPP was first dissolved in toluene (concentration 1 mg ml⁻¹), sonicated for 10 s, then added to the base component of a 2-part PDMS silicone (Silgard 184, Dow-Corning, Midland, MI, USA), for a final concentration of 0.05 mass per cent (the ratio of the mass of the PtTFPP to the mass of the total mixture, multiplied by 100). The toluene was then evaporated off using a vacuum centrifuge (Eppendorf Vacufuge Concentrator, Eppendorf Canada, Mississauga, ON, Canada). The PtTFPP and PDMS base was combined with the PDMS catalyst in a ratio of 10:1 base:catalyst and thoroughly mixed. This mixture was placed under a vacuum to remove any air bubbles incorporated during mixing, then a droplet was placed in the middle of a clean microscope slide. The droplet was left to spread over the surface of the slide, before the slide was placed on a temperature-controlled Peltier plate set to 40°C overnight. The final thickness of the PDMS film was 0.09 mm, measured using digital Vernier callipers.

To measure any change in P_O2 within a spittle mass containing a nymph, a square frame with internal dimensions of 17×17 mm, wall thickness of 3 mm and height of 0.8 mm was 3D printed from ABS plastic. This was laid flat in the middle of the PDMS film. A nymph was placed into the square frame and its spittle mass was deposited on top of it, submerging it completely. A coverslip was then placed over the frame, sandwiching the foam and spittlebug inside the frame. A 1 mm gap cut into the middle of one of the square's sides allowed excess foam to escape from beneath the coverslip. The frame and coverslip held the spittlebug securely against the PDMS film within the foam, while preventing it from being crushed by the weight of the coverslip. The microscope slide was then placed on the stage of an inverted epifluorescence light microscope (IX73, Olympus, Tokyo, Japan). Excitation light (390 nm peak) from an X-Cite 120 LED light source (Excelitas Canada Inc., Vaudreuil-Dorion, QC, Canada) was used to excite the PtTFPP, while the emitted light was passed through a 645 nm (75 nm bandwidth) emission filter (ET645/75 m, Chroma Technology Corp., Bellows Falls, VT, USA) before being imaged using an Orca-Flash 4.0 LT CMOS camera (Hamamatsu, Hamamatsu City, Japan) using CellSens imaging software (Olympus, Tokyo, Japan). Images were taken at 10 min intervals. The fluorescence intensity of the PDMS film over time was measured at five locations in each image: the air bubble associated with the tip of the spittlebug's abdomen and four other, non-adjacent, air bubbles. ImageJ was used to calculate the average fluorescence intensity within each of these areas. Increases in fluorescence intensity within each bubble, associated with decreases in P_O2, were quantified by dividing the fluorescence intensity within each bubble at each time point by the initial fluorescence intensity within the bubble at time zero.

All V̇_O2 and V̇_CO2 values used in the calculation of RQ were averaged from the last 1.75 h of the respirometry trials to reduce any influence of active movement on the measurements. RQs were calculated by dividing the mean, mass-specific V̇_CO2 by the mean, mass-specific V̇_O2. The V̇_CO2 of nymphs feeding on xylem was a grand mean averaged from all active foam production periods, and V̇_CO2 for walking was the grand mean of all continuous walking periods during the 2 h trials. All mass, V̇_O2 and V̇_CO2 values were log-transformed before linear regressions were applied to them.

Means, standard deviations, linear regressions and Johnson–Neyman tests were calculated in Microsoft Excel, and all other statistical tests and graphs were made using R (version 3.3.2; www.r-project.org) inside RStudio (version 1.1.383).

There was no relationship found between P_O2 and depth within the masses of foam measured in the field (F_1,304=0.5368, P=0.4643) (Fig. 2). The mean (±s.d.) P_O2 within the spittle was 19.97±1.17 kPa, and did not vary from the edge to centre of the spittle (n=306). This P_O2 was not significantly different to the atmospheric level measured directly outside the spittle mass.

For all newly secreted masses of spittle, the P_O2 was very slightly lower inside the foam than in the air outside it. This difference was statistically significant at the beginning of the trial (paired t-test, t₁₃=6.65, P=1.59e−05). At the beginning of the 2 h measurement period, the mean (±s.d.) P_O2 inside the freshly produced spittle was 20.1±0.42 kPa (n=14), and continued to increase to an average of 20.94±0.39 kPa (n=14) by the end of the 2 h.

The P_O2 within bubble chambers formed by moulting final instar spittlebugs was always lower than that in the outside air (paired t-test, t₂₂=9.4581, P=3.294e−09) by an average of 1.91±0.97 kPa (n=23). It was observed that the P_O2 would slowly rise during the measurement in association with the gradual drying out of the wet foam covering the bubble chamber. When the bubble ultimately popped, the P_O2 immediately increased to the ambient level.

In order to calculate the O₂ consumption of individuals, the body density of both spittlebug nymphs and adults was determined. Philaenus spumarius nymphs have an average density of 1.11±0.0145 g cm⁻³ (n=3) and adults have an average density of 1.05±0.02 g cm⁻³ (n=3).

After log transformation, the nymphs' V̇_CO2 (µl h⁻¹) followed the equation log(V̇_CO2)=0.845log(M)+0.0725, and V̇_O2 (µl h⁻¹) followed the equation log(V̇_O2)=0.7649log(M)+0.1872, where M is nymph mass in mg. Using the Johnson–Neyman technique, spittlebug nymph V̇_CO2 and V̇_O2 while at rest were not found to be statistically different among slopes (F_1,85=0.2124, P=0.6461) but were found to be statistically different among elevations (F_1,86=5.55, P=0.0207; Fig. 3). The resting V̇_CO2 and V̇_O2 of adults, however, did not statistically differ among slopes (F_1,37=0.581, P=0.451) or elevations (F_1,38=0.467, P=0.5; Fig. 4). The V̇_CO2 of adult froghoppers followed the equation log(V̇_CO2)=1.39log(M)−0.5, and V̇_O2 followed the equation log(V̇_O2)=1.095log(M)−0.209. Using these measured V̇_CO2 and V̇_O2 rates, the RQ for P. spumarius was calculated to be 0.92 for nymphs and 0.95 for adults.

The V̇_CO2 of spittlebugs during locomotion followed the equation log(V̇_CO2)=0.464log(M)+0.478, whereas the V̇_CO2 of spittlebugs while actively feeding on xylem and excreting spittle followed the equation log(V̇_CO2)=0.342log(M)+0.6695. The V̇_CO2 of spittlebugs while making spittle was compared with production while walking using the Johnson–Neyman technique and they were found to follow the same slope (F_1,26=0.11, P=0.743) but had different elevations (F_1,27=5.26, P=0.03). V̇_CO2 during spittle production and feeding was on average 20±0.1% higher than when the same individual was walking (n=10, 3 outliers removed; Fig. 5). These values were compared with the RMR values, which showed that the slopes for RMR and metabolic rate during xylem feeding did not differ (F_1,30=4.078, P=0.054), but their elevations did differ (F_1,31=20.41, P<0.001; Fig. 5). RMR did not differ in slope (F_1,26=2.82, P=0.1) or elevation (F_1,27=1.679, P=0.21) from walking metabolic rate.

After a spittlebug had made a foam mass that completely covered it, the respirometry chamber was tapped sharply with metal forceps. Immediately after each knock, the nymph was observed to withdraw completely into the spittle. This movement was accompanied by an immediate drop in V̇_CO2 close to the chamber's baseline before returning back to the average production value that the nymph had had during feeding and spittle production. This drop in V̇_CO2 lasted an average of 46.5±7.84 s (n=17; Fig. 6). There was no correlation between the length of time CO₂ dropped and the mass of the spittlebug nymph (F_1,15=0.557, P=0.47).

The first response of the spittlebugs forcibly submerged in the foam was to consolidate multiple smaller bubbles in the foam into a single large bubble (Fig. 7). This occurred as the mobile tip of the spittlebug's abdomen moved from bubble to bubble, breaking the surface tension between them, combining them into a single volume. For the duration of the measurement, the tip of the spittlebug’s abdomen remained within this air bubble. Over time, the level of fluorescence increased in the PDMS film in contact with this large bubble, but remained constant in all other regions of the foam (Fig. 8). A rise in fluorescence is consistent with a decreased P_O2 in the large air bubble, and a concomitant decrease in O₂-quenched fluorescence.

The absence of a sub-atmospheric P_O2 within spittlebug foam, as well as the continuous V̇_CO2 measured from feeding spittlebugs using flow-through respirometry, shows definitively that P. spumarius snorkel: they breathe by extending the tip of their abdomen out of the foam mass, thereby bringing atmospheric air into contact with their ventral grove and abdominal spiracles. When a spittlebug retracts the tip of its abdomen into its foam mass, the measured V̇_CO2 drops immediately. Newly produced foam has a P_O2 that is very slightly below atmospheric levels (<1 kPa), presumably as the incorporated air has just been expelled from the spittlebug's ventral groove. But this P_O2 gradually rises over time to ambient levels, indicating that it is not subsequently being used for respiration. Even when startled, the insects would only submerge completely within their foam for 46.5 s, on average, and never longer than 70 s. This suggests that they do not attempt to access O₂ from the bubbles of air trapped within the foam mass to prolong their submergence. Taken together, this evidence disproves the theory that spittlebugs use their foam as a ‘respiratory organ’ (Turner, 1994). That a wet foam is likely to offer a significant barrier to the uptake of atmospheric O₂, despite the presence of air bubbles, has also been demonstrated in the aerial foam nests of frogs (Seymour and Loveridge, 1994). This same result can be seen clearly in Figs 7 and 8. After 125 min, only the air bubble that is in contact with the spittlebug's ventral groove shows any decrease in P_O2, as indicated by the increase in fluorescence. There is no substantial diffusion of O₂ from either the air bubbles or fluid in the surrounding foam. Thus, if an individual were to become trapped within its own spittle, it could use the tip of its abdomen to respire the O₂ within the foam's air bubbles, but this appears to be an action of last resort.

The one period when P. spumarius nymphs must remain submerged within their spittle occurs during the moult from final instar into adult. During this transition, the spittlebug produces a large single bubble of air in the centre of the foam (Fig. 1C). How they produce this chamber has not been determined, but it is entirely plausible that the spittlebug uses the same bubble-consolidation process as was observed when spittlebugs were forcibly submerged in their foam: the tip of their abdomen can break the surface tension of the spittle film around the bubbles, causing them to merge. Repeating this process within the centre of the spittle mass would consolidate the many small bubbles into a single large bubble chamber. However, the adult froghopper that emerges from the exuviae of the final instar must remain sealed within this bubble chamber for some period of time, as it possesses neither the flexible abdomen nor the ventral groove needed to break through the foam and access atmospheric O₂. As a result, the P_O2 inside the bubble chamber was on average 1.9 kPa lower than that of the ambient atmosphere during the first 15 min of measurement. Interestingly, all bubble chambers showed a gradual increase in P_O2 during the 2 h measurement. This can be attributed to a gradual increase in the conductance of the spittle mass. Following moulting, the adult froghopper no longer feeds or adds excreted xylem fluid to the spittle mass. As a result, the spittle dries out, reducing the thickness of the liquid – the principal barrier to O₂ diffusion through the foam. By the time the adult froghopper is ready to emerge, the foam usually has an almost powdery consistency and is probably quite permeable. From these measurements it appears that final instar spittlebugs and newly moulted adults are unlikely to ever experience more than a very minor hypoxia, even inside a completely enclosed bubble chamber.

From the resting V̇_O2 and V̇_CO2 values measured in this study, it can be seen that the nymphs and adults of P. spumarius have a RQ of 0.92 and 0.95, respectively, a finding that is entirely reasonable for an insect that feeds exclusively on the sugars and amino acids in plant sap. The resting V̇_O2 of spittlebug nymphs measured in this study also agrees closely with previous measurements on P. spumarius nymphs. Wiegert (1964) used a manometric respirometry technique and found that nymph V̇_O2 was on average 9.2 μl h⁻¹ for fifth instars at 25°C, compared with the 9.8 μl h⁻¹ calculated for an 11 mg (fifth instar) nymph in this study.

Feeding on xylem sap is assumed to be energetically expensive because of the presence of negative pressures within the xylem (Kim, 2013; Novotny and Wilson, 1997; Raven, 1983). Despite the expected cost of this feeding strategy, previous studies on P. spumarius have concluded that this species is capable of rapidly extracting large volumes of xylem sap under tensions exceeding 10 bar (1 MPa), as measured using a Scholander pressure bomb (Malone et al., 1999). Other studies have also found that the high fluid excretion rates of xylem-feeding sharpshooters (Hordinia circellata) are unaffected by manipulations that should increase the hydrostatic tension within the xylem of their host plant by 1 bar (Mittler, 1967). However, feeding rates in an adult leaf hopper (Homalodisca coagulata) decreased dramatically once xylem tension exceeded 1 MPa (Andersen et al., 1992). These findings are curious and suggest two possible explanations: either these bugs are capable of routinely producing pressures in excess of −1 MPa with their cibarial pumps, regardless of the metabolic cost, or the xylem pressures they experience while feeding are far lower than measured using a pressure bomb (Kim, 2013). On mechanical grounds it has been argued that a cibarial pump should not be able to produce a suction much greater than 0.3 MPa (Raven, 1983), supporting the latter possibility.

By measuring the metabolic rate of a spittle bug as it feeds on xylem, however, we have quantified the energetic cost of this activity; feeding P. spumarius spittlebugs had metabolic rates that were on average only 20% higher than their RMR. In a 9 mg insect, this is equivalent to a net feeding cost of ∼1.82 µl O₂ h⁻¹, or 0.0378 J h⁻¹ (assuming a RQ of 0.92 and an oxyjoule equivalent of 20.75 J ml⁻¹ O₂; Lighton, 2008). The rate of fluid excretion during these metabolic rate measurements was also high, with the spittlebug being completely covered by a foam mass within 23.22±2.8 min (n=9). This suggests that the cost of extracting xylem sap is low.

A low feeding cost associated with xylem extraction is also reasonable considering the very low sugar content of xylem sap. Assuming that xylem sap contains 0.3 g glucose l⁻¹ (1.66 mol m⁻³ of reducing sugars; Raven, 1983) and the spittlebug had an excretion rate of 72 μl xylem h⁻¹ (20 nl s⁻¹; Malone et al., 1999), then a feeding spittlebug should be able to obtain a maximum of 21.6 μg sugar h⁻¹ if 100% of the sugar was extracted. Based on the insect's total metabolic rate of 0.23 J h⁻¹, then extracting this quantity of xylem would provide approximately 1.5 times the insect's total metabolic rate requirements.

Furthermore, from the net increase in metabolic rate of a feeding spittlebug, it is possible to estimate the cibarial pump pressure. If a feeding spittlebug has a net feeding cost (xylem pumping cost) of 0.0378 J h⁻¹ (1.05×10⁻⁵ J s⁻¹ or W) and a volumetric flow of xylem of 20 nl s⁻¹ (2×10⁻¹¹ m³ s⁻¹), then the cibarial pump pressure would be approximately −0.53 MPa. However, this calculation neglects the less than 100% efficiency of muscle. Assuming that muscle efficiency in insects lies between 10% and 50%, depending on the muscle type [skeletal muscle 50% (Pybus and Tregear, 1973), flight muscle 10% (Ellington, 1985)], this would reduce the cibarial pump pressure proportionately, to −0.26 to −0.05 MPa. For a spittlebug to draw up xylem sap, the pressure produced by their cibarial pump must be must be more negative than that of the sap in the xylem vessel. Pump pressure must also be sufficiently low to overcome the pressure drop that is attributable to the resistance of the spittlebug's stylets (which may or may not be substantial). This means that the xylem pressure of the sheep sorrel plants in these experiments must be less negative than the pump pressures calculated here. This would place the calculated pressures towards the lower end of the range reported from maize xylem (−0.7 to −0.1 MPa), as measured using both the Scholander pressure bomb and implanted pressure probe techniques (Wei et al., 1999). They are also below the theoretical maximum cibarial pump pressure of −0.3 MPa as estimated by Raven (1983). Thus, it remains to be seen whether the low pressures calculated here are representative of the xylem pressures experienced by feeding spittlebugs, or whether these insects are indeed capable of feeding from xylem with tensions exceeding 1 MPa, as has been suggested (Malone et al., 1999).

Besides P. spumarius, the most comprehensive analysis on the cost of fluid feeding in insects has been undertaken using Rhodnius prolixus, a triatomid hemipteran that sucks blood from a vertebrate host, also using a cibarial pump. So, how does the cost of xylem feeding compare with the cost of blood feeding? Fluid feeding was found to increase the metabolic rate of R. prolixus by up to 17-fold its pre-feeding metabolic rate (Leis et al., 2016). This dramatic increase in metabolic rate led Leis et al. (2016) to conclude that haematophagy is metabolically very expensive. But this conclusion ignores the fact that blood-feeding insects, and in particular R. prolixus, significantly depress their metabolic rate in-between meals (Bradley et al., 2003). Thus, given a low enough pre-feeding RMR, the factorial increase to even a routine metabolic rate (by the standards of other insects) will appear exceedingly high. Comparing the absolute mass-specific metabolic rate before and during feeding for both P. spumarius and R. prolixus reveals this difference. Pre-feeding R. prolixus have a metabolic rate of 0.00296 J h⁻¹ mg⁻¹ (Leis et al., 2016) compared with that of P. spumarius of 0.021 J h⁻¹ mg⁻¹ (calculated from the V̇_O2 of a 9 mg nymph using a RQ of 0.92 and the same oxyjoule equivalent of 20.75 J ml⁻¹ O₂). Thus, a pre-feeding spittlebug has a mass-specific metabolic rate that is 7.1 times higher than that of R. prolixus. In comparison, R. prolixus feeding on saline solution (as a reasonable proxy for xylem sap, but without the negative tension) has a mass-specific metabolic rate of 0.028 J h⁻¹ mg⁻¹ (Leis et al., 2016) compared with 0.025 J h⁻¹ mg⁻¹ in the xylem-feeding P. spumarius. This reveals that the mass-specific metabolic rates of these two bugs during fluid feeding are very similar, with the spittlebug's metabolic rate actually being slightly lower. However, differences in cibarial pump morphology, feeding rates, stylet resistance, fluid pressure and viscosity would all factor into interpreting these costs. For example, average pumping frequency in R. prolixus is 6 Hz (Smith, 1979), compared with the average of 1.3 Hz for a spittlebug measured in this study. The volume of fluid per pump is also much higher in R. prolixus [57.8 nl pump⁻¹ in R. prolixus (Smith, 1979) versus 11.8 nl pump⁻¹ in P. spumarius (Malone et al., 1999)]. Thus, understanding the differences in the form and function of cibarial pumps has the potential to reveal much about the limitations of fluid feeding in these remarkable insects.

This study shows that spittlebug nymphs breathe atmospheric O₂ continuously by extending the tip of their abdomen out of the spittle mass. This allows these insects to continuously take up O₂ and release CO₂ while feeding on xylem. The increase in metabolic rate that accompanied xylem feeding is modest compared with their metabolic rate during rest and locomotion. Estimations of xylem pressure from the cost of feeding appear reasonable, and certainly within the range of xylem tensions measured in other studies. However, further studies should be done to investigate the metabolic rate, and rates of xylem excretion, of spittlebugs feeding on known xylem tensions to better understand the true metabolic costs associated with extracting this sap from plants.

We thank the two reviewers for their constructive criticism of this paper. Thanks also to Hsin-Yun Tsai and Russ Algar for their assistance with the vacuum centrifuge, and Karen Needham for species identification.