Thrips' responses to thermogenic associated signals in a cycad pollination system: the interplay of temperature, light, humidity and cone volatiles



  1. Most functions postulated for thermogenesis in plant reproductive organs are associated with pollination and pollinator activity. During thermogenesis other chemical changes occur, and these also may affect pollinator activity.

  2. We address how cone thermogenesis and the associated chemical emissions influence pollinator behaviour in the obligate mutualist Macrozamia lucida and M. macleayi cycad – Cycadothrips chadwicki thrips pollination system. Cones of these dioecious gymnosperms have a diel, mid-day metabolic burst that is associated with increased cone temperatures, volatile emissions (primarily β-myrcene), and CO2 and water vapour emissions. Concurrently, thrips leave cones en masse and then return to cones later in the day. We investigated the effects of these cues, individually and in combinations, on their potential to induce thrips to leave cones in the dark and light with a suite of Y-tube behavioural experiments.

  3. The results suggest that ambient light, and high cone temperatures, humidity, and β-myrcene levels, but not CO2, each induce Cycadothrips to leave cones. At typical overnight temperatures (14 °C) thrips were relatively inactive and negatively phototactic. At typical daytime ambient temperatures, 22–26 °C (lower than typical thermogenic temperatures), thrips were active and positively phototactic. Thrips moved away from or avoided thermogenic temperatures and high concentrations of β-myrcene, and they preferred dry (<8% RH) to humidified (~88% RH) air.

  4. Whereas several variables individually induce thrips to leave a cone's darker interior towards the daytime light, these signals are presented simultaneously during thermogenesis. There thus seems to be enhanced or redundant signalling that enforces thrips departure from pollen cones thus increasing the chances of vectoring pollen to ovulate cones. While high temperature appears necessary to mediate thrips movement, its concurrence with other plant signals that produce similar responses, suggests a dynamic multimodal signalling system that operates as a functional unit.


Plants use a diversity of signals to manipulate animal mutualists, predators and parasites by attracting, retaining or repelling them in the context of their particular relationship or phase of the relationship (Schiestl et al. 1997; Schiestl & Ayasse 2001; Raguso 2004, 2009). These signals include well-studied visual attractants (shapes and colours, including hue, saturation and brightness), odourants emitted by the plants, and toxins within specific tissues. These multiple signal types can work in redundant, complementary, or synergistic ways (Raguso & Willis 2002; Raguso 2004) that may differentially affect pollinators over herbivores (Theis & Raguso 2005; Theis, Lerdau & Raguso 2007; Halitschke et al. 2008; Kessler, Gase & Baldwin 2008; Kessler & Halitschke 2009).

Plant attractiveness may decrease simply by the reduction of resources during maturation or after pollination; visual or volatile cues may fade, repellents may be emitted, or rewards may diminish (e.g., Schiestl & Ayasse 2001). In plants pollinated by a host specific insect that feeds only on that plant as adults and larvae, pollination success relies on the adult leaving its host at appropriate times. Such departures may be contrary to the immediate requirements of the pollinator. In addition to the plant's status, the pollinator's physiological state may be important as in some fig-fig wasp pollination systems (Anstett, Hossaert-McKey & Kjellberg 1997).

Thermogenesis has been detected in the reproductive organs of many plant species within several angiosperm families (Azuma, Thien & Kawano 1999; Dieringer et al. 1999; Thien, Azuma & Kawano 2000; Seymour, White & Gibernau 2003) and within gymnosperms, including many cycad species (Tang 1987a, 1993; Donaldson 1997; Seymour, Terry & Roemer 2004; Terry et al. 2004a; Roemer et al. 2005, 2012; Roemer, Terry & Walter 2008; Suinyuy, Donaldson & Johnson 2013) and some pines (Takács et al. 2008). Several functions have been proposed for this energetically expensive activity (Moodie 1976; Búrquez, Sarukh & Pedroza 1987; Seymour 1997; Seymour & Schultze-Motel 1998; Seymour, Schultze-Motel & Lamprecht 1998; Ervik & Barfod 1999; Patiño, Grace & Bänziger 2000; Barthlott et al. 2009), most of them associated with pollinator activity. These include the attraction of pollinators on the one hand, the stimulation of their dispersal on the other, and the provision of thermal energy to either larvae or adults while they are enclosed in the plants' reproductive structures. These interpretations of the effects of this energetically costly temperature elevation on pollinator behaviour have been derived from only a few species of plants (see Stensmyr et al. 2002; Seymour, White & Gibernau 2003, 2009; Angioy et al. 2004; Seymour & Matthews 2006), none of them gymnosperms.

Cycads are dioecious gymnosperms occurring in tropical, subtropical and warm temperate regions (Norstog & Nicholls 1997; Jones 2002). Many species have a single specialist pollinator that depends entirely on pollen cones for oviposition, larval feeding and development, and adult feeding (Norstog, Stevenson & Niklas 1986; Tang 1987b; Norstog & Fawcett 1989; Norstog, Fawcett & Vovides 1992; Donaldson 1997; Stevenson, Norstog & Fawcett 1998; Terry 2001). Ovulate cones, by contrast, provide essentially nothing, with the possible exception of minute quantities of pollen droplet liquid (Stevenson, Norstog & Fawcett 1998; Tang 2004; Terry et al. 2005). Movement of pollinators out of the pollen cones is critical to cycad pollination and reproduction, yet little is known about the signals that induce pollinators to leave pollen cones.

The thrips Cycadothrips chadwicki Mound (Thysanoptera: Aeolothripidae) is the sole pollinator of several Macrozamia cycad species, including M. lucida L.A.S. Johnson and M. macleayi Miq. During their pollination period, pollen and ovulate cones have a mid-day diel thermogenic event lasting several hours. Cycadothrips leave cones en masse during these events (Fig. 1), and later, in the day, they return to cones, most to pollen cones but some enter ovulate cones (Terry et al. 2004b, 2005). Cones provide a dynamically changing intersporophyll micro-environment (see Fig. 2): cone temperatures as low as 9–10 °C overnight and as high as 40 °C during the mid-day peak of thermogenesis; changes in light levels; cone emissions of β-myrcene that vary from barely detectable to concentrations several thousandfold higher (Terry et al. 2004b); and CO2 and humidity changes from near ambient air levels to as high as 6% CO2 and 100% RH (in gas exiting a flow-through respirometry chamber containing a thermogenic cone (see Appendix S1, in Supporting Information)).

Figure 1.

Cycadothrips chadwicki leaving a thermogenic pollen cone of Macrozamia lucida (photo credit Irene Terry).

Figure 2.

Illustration of the changes in key signals occurring in pollinating cones of Macrozamia lucida and M. macleayi overnight and during a mid-day thermogenic event when thrips exit cones. Non thermogenic cones track the ambient conditions. Experiment numbers match those in the text that test that particular signal. See Appendix S1, for CO2 and ß-myrcene estimations.

Cycadothrips thus experience dramatic changes in physicochemical conditions during these diel thermogenic episodes, but it is not clear which of these multiple potential cues affects their behaviour. Temperature clearly affects the behaviour of ectotherms, but the effect of temperature alone on Thrips' movement within cones has been difficult to test. That is, if pollen cones are heated externally to determine the effects of high temperatures alone, without volatiles, cones will immediately begin a thermally triggered metabolic rate increase with rises in CO2, RH and volatiles. Because of such constraints, we performed a series of Y-tube-choice experiments to test CycadoThrips' behavioural responses to temperature, ß-myrcene, and different levels of humidity and CO2. Our expectation was that each signal would be detected by and would stimulate thrips to leave cones, or at least that different signals would not stimulate contradictory behaviours. Our predictions for CO2 and relative humidity were based on prior research on other arthropods responding to these cues. Our focus on ß-myrcene extends earlier research in which thrips responded to β-myrcene, the component comprising over 90% of the volatile emissions, in a concentration-dependent, push-pull manner (Terry et al. 2007b).

Interest in complex signalling (containing multicomponent or multimodal signals, or both), among animals and its evolution has increased in the last few decades, building on earlier research which focused on the effects of single signals (see review by Hebets & Papaj 2005). Plant communication with animal mutualists or predators can be considered under the same potential rubric. We examined how the different biotic and abiotic signals in the Macrozamia-Cycadothrips pollination system operate as a functional whole, as outlined by Hebets & Papaj (2005). Specifically, we address whether a signal is necessary and/or sufficient to elicit a response, and for some of the signals, whether the presence of one signal influences the pollinator response to another signal. We then explore ways in which this cycad system relates to different hypotheses of complex signal functioning.

Materials and methods

Test species

Cycadothrips chadwicki were obtained from Macrozamia lucida and M. macleayi pollen cones from field populations. Both cycad species co-occur in southeastern Queensland (Qld) near Brisbane, Australia, in Brisbane Forest Park (Terry et al. 2004b) and Moggill Conservation Park. Although these Macrozamia species have been placed into separate Macrozamia subsections, Parazamia and Macrozamia respectively, hybrids between them (based on morphology) are present where these co-occur. Both species have C. chadwicki as their sole, obligate pollinator, and both have similar thermogenic events and volatile emissions (Terry et al. 2004b). Generally, coning plants can be found from mid-October to late November, and only a small fraction of plants within a population are reproductive in any particular year. Pollen cones can be excised and maintained in the laboratory for several weeks during their pollination phase, where their diel thermogenic behaviour persists (Tang 1987a; Roemer et al. 2012). Due to the short pollination season, and thrips being present only during this season, studies were conducted over four pollination seasons. Presumably thrips spend the rest of the year as pupae in the soil.

Cones containing Cycadothrips were kept at temperatures similar to natural conditions (Roemer, Terry & Walter 2008); that is, in a dark, temperature controlled environator (14 °C) overnight and then in a laboratory (~23 °C) near a window during the day. Cones would begin thermogenesis after ambient temperatures warm them to ~21 °C (Roemer, Terry & Walter 2008), and thrips would depart during thermogenesis. Thrips then flew to a window pane where they constantly moved around until they returned to the cone late in the afternoon. Thrips of each sex were collected periodically from the window pane in preparation for each behavioural test.

Test procedures and thrips preparation

Almost all tests were performed with directional signals that are consistent with what the insects experience in situ. For example, a thrips inside a dark thermogenic cone would be exposed to an exterior lighting source. A few tests with altered directional signals were performed to compare the strengths of competing cues. Because tests were designed to deconstruct and tease apart the effects of different signals, there were no positive controls used in any of these experiments, and treatments are only compared with air controls or against other signals. Animals were wild collected and thus were not naïve to any of the signals tested. Although no specific experiments tested for circadian rhythm in thrips behaviour, the results from a wide range of test times throughout the day (primarily 11:30–18:00 h, but as early as 0830 h and as late as 2030 h) did not suggest a change in behaviour due to time of day.

Except for preliminary tests in straight glass tubes (See Appendix S2), all tests were performed in a Y-tube apparatus. The Y-tube had an internal diameter of 1 cm, and segment lengths of: base, 3 cm; stem, 7 cm; and arms, 10 cm. The stem and arms were clear glass, while the base was brushed glass and was always covered with a dark cloth. Ground glass filters located between the base and the vacuum line and between the Y-tube arm tips and the treatment vessels prevented thrips from escaping from the apparatus.

Experiments were performed with a control arm (air at room temperature and humidity unless otherwise noted) and a treatment arm carrying the modified air (increased temperature, relative humidity, β-myrcene or CO2) to the junction of the Y-tube's arms where the two flows combined. For all tests, except those involving CO2 and humidified air, air flow was produced with a vacuum line attached to the base of the Y-tube. The resulting pressure gradient pulled room air successively though activated charcoal filters, the treatment vessels (one sham in the control arm and one active in the treatment arm), and the Y-tube's arms, stem and base. A calibrated in-line valve controlled the stem/base air flow at ~70–90 mL min−1 with the flow in each arm controlled at ~40 mL min−1 per arm by calibrated rotameters (Anavac, Aalborg low volume gas flowmeter, Monsey, NY, USA) with adjustable in-line valves. Preliminary tests (see Appendix S2, Tables S1 and S2) examined the effects of air flow (no flow vs. ~90 mL min−1) on Thrips' responses in the cold, and there was no significant effect; therefore, all tests were conducted using flowing air.

Prior to each experiment, male or female thrips, ~10–15 per replicate, were placed into an Eppendorf tube and allowed to acclimate to the ambient test conditions (light or dark, and ambient temperature) for at least 30 min because preliminary tests showed an effect of light versus dark preconditioning at 14 °C, although there was no effect at 22 °C (see Appendix S2 Tables S3 and S4). To initiate a test, we tapped acclimated thrips into the Y-tube base, which was immediately connected to the Y-tube stem. Thrips were then allowed to move to other segments of the Y-tube (stem, test arm or control arm) for five minutes, at which time the number in each segment was recorded. Most experiments were performed with eight replicates at each factor level, with four replicates of each sex, resulting in a minimum of 80 thrips for each test. At the completion of each replicate, thrips were removed and a new set of thrips was used for the next replication. Tests with air in both arms were performed intermittently to check for bias towards one Y-tube arm or the other. In addition, the treatment arm was alternated frequently between the left and right position, except in the heated arm tests, in which the heated arm was in a fixed position.

Lighting conditions

Tests were performed in an unlit, windowless interior laboratory with a fluorescent lamp providing the only light to the Y- tube. The lamp was centred on the Y arms' central axis and located at horizontal and vertical distances from the Y-tube arms' tips of 24 and 27 cm, respectively. This resulted in Photon Flux Density light levels (measured with a Skye SKP 200/215 PAR quantum sensor; Skye Instruments Ltd, Llandrindod Wells, Powys, UK) of: tips of the Y-tube's arms, 10 μmol m−2 s−1 (symmetric at both tips); junction of arms and stem, 8·7 μmol m−2 s−1; and, junction of stem and base, 6·6 μmol m−2 s−1. Those levels are comparable to the lighting experienced at mid-day by understory field cones. Lamp lighting was present for all tests except where noted. In all tests, the Y-tube's base was kept dark with layers of dark towelling. For many tests, we used ‘dark- to light-changeover’ lighting conditions that involved 5 min of dark conditions after which the Thrips' locations were noted, followed by 5 min of lighted conditions at which time the new distribution of thrips was recorded.

Temperature plus lighting experiments

Each of the six experiments performed is given a number (1–6) that refers to the attribute being tested as outlined in Fig. 2. The first three, Experiments 1–3, were designed to examine pollinator movement under different temperature and lighting conditions. The uniform temperature tests (Experiment 1) examined thrips movement in the dark and light at different ambient temperatures ranging from typical overnight to typical daytime temperatures. The heated base and arm tests (Experiments 2 and 3) examined whether thrips would move away from, or avoid, typical thermogenic temperatures (~28–36 °C) and go towards cooler temperatures.

Experiment 1

Uniform temperature tests were performed at different room ambient temperatures in a temperature controlled room, with the room and all segments of the Y-tube at either 14, 18, 22 or 26 °C (±~0·8 °C), which encompasses the daily range of average air temperatures in the field (Terry et al. 2004b). Each test was conducted using the ‘dark-to-light changeover’ protocol noted above to determine how thrips moved at different uniform temperatures in the dark, initially, and then when exposed to light.

Experiment 2

Heated base tests were performed under conditions in which the room and most of the Y-tube were kept at 22·3 °C (±0·6 °C), but with the Y-tube base heated separately (26–36 °C) by an adjustable hot plate that controlled the temperature of the base. Since thrips are highly attracted to light at typical daytime temperatures (see uniform temperature test results below), the heated base tests were performed only in the dark to avoid presenting a strong attractive light signal. The heated base temperatures were measured in the centre of the heated tube's lumen at the junction of the base and stem with a type K thermocouple using a portable data acquisition system (Fluke Model 54II, Fluke, Inc., Everett, WA, USA).

Experiment 3

Heated arm tests were performed with the room and Y-tube kept at 22·3 °C (±0·6 °C), but one of the Y-tube arms was heated separately with the hot plate adjusted to a series of set temperatures reaching up to 38 °C. The heated arm temperatures were measured in the centre of the heated arm's lumen, just upstream of the junction of the arms, which is where thrips respond to conditions in one or the other Y-tube arm. With this arrangement: (i) thrips in the base and stem are exposed to an intermediate temperature resulting from mixing of the air from the heated and control arms; and (ii) upstream positions inside the heated arm have higher temperatures than at the junction. These tests were performed under several light conditions: all Y-tube segments darkened; all segments lighted (except the base, which was always covered); only the heated arm lighted; and only the unheated arm lighted. These tests allow comparisons of the movement of thrips towards or away from a heated source at different temperatures which was either lighted or darkened. Base and arm temperature measurements were made both before and after several replicate tests by inserting the thermocouples into the sampled locations while air was flowing at the test rate.

Experiments with other signals

Experiment 4

β-myrcene and temperature experiments: Previous Y-tube tests in lighted conditions at ~24 °C demonstrated that thrips were slightly attracted to low concentrations of β-myrcene but clearly avoided high concentrations (≤5 μL and 100 μL in castor oil, respectively) (Terry et al. 2007b). Because thrips experience different combinations of temperatures, volatiles, and lighting conditions during the day, we compared the responses of thrips to β-myrcene (Sigma, ≥ 90% purity, see Appendix S3) at concentrations of either 5, 100, or 200 μL mixed in six drops of castor oil in one treatment arm of the Y-tube versus a castor oil control arm, and at uniform ambient temperatures of 14, 18, 22, and 26 °C. These tests were conducted under the dark-to-light changeover conditions.

Experiments 5 and 6

CO2 and humidity: Air in the control arm was supplied by pressurized instrument grade air (BOC, Brisbane, Australia), and the CO2 arm was supplied either directly from a pressurized, calibrated CO2 tank, at 2·06% or 5·05% ± 0·1% (BOC), or from a mixture of CO2 from a pure CO2 tank and air from the instrument grade air tank to produce 2, 10 and 20% CO2. Flow rates were measured and controlled with the adjustable valve rotameters. In these tests, the activated charcoal filters and treatment vessels were not used, and the tips of the arms were covered with fine nylon mesh to prevent escapes. These tests were conducted using either light only or dark- to light-changeover conditions.

In the relative humidity tests, air was supplied to both arms from pressurized cylinders (either instrument grade air or CO2). All tanks had very low humidity at room temperature (air tank, <5% RH; CO2 tanks, <8% RH), as measured by HOBO RH Pro v2 sensors (Onset Computer Corp., Bourne, MA, USA). The relative humidity of the air supplied to a treatment arm was increased by bubbling the air through water-filled Erlenmeyer flasks before it entered the Y-tube arm, yielding a relative humidity of ~85–88%. To ensure that the air everywhere in the test arm was at the desired humidity level, the Y-tube system was allowed to equilibrate with humidified air for five minutes before initiating each test. These tests were performed at a room temperature of 24 °C under lighted conditions only.

Statistical analysis

For each Y-tube experiment, several metrics were examined using logistic regression models (The R Foundation for Statistical Computing, version 2.15.1, glm, binomial logit link models) or t-tests: the proportion of thrips that remained in the base and did not move to another segment (any damaged thrips during transfer, generally <1%, were not included); the proportion of thrips that moved into the arms; and the proportion that moved into the treatment arm (either of the total thrips that moved into both arms or of the total thrips in the test). Several covariates, including thrips gender, lights on or off, β-myrcene level, temperature, CO2 or relative humidity, were tested for their effects on thrips responses, i.e., movement in the Y-tube. The logistic binomial models were checked for overdispersion (residual deviance/residual d.f. ~1) and then the best model was selected on the basis of the lowest Akaike's Information Criterion score (AIC) and the significance of the goodness of fit z value for each parameter. For parametric statistics, the data, or transformed data, were tested for normality (Shapiro–Wilk test) and for variance equality (Statistix 9.0, Tallahassee, Florida, USA) assumptions. Because of the number of different signals tested, and to reduce the familywise error rate, we used an overall probability level of 0·01 as the minimum value to accept a factor or a factor level as having a significant effect (Glantz 2005).


Experiment 1: Thrips responses to light at different ambient temperatures

Results of the dark-to-light changeover tests (in which data from both arms are combined because both arms are at the same temperature and lighting) indicate that at a low ambient temperature (14 °C) thrips favour the dark, for most thrips remained in the Y-tube base during the dark period (Fig. 3; see Appendix S4, Table S5 and Fig. S2). When the light was turned on, most of the thrips that had moved out of the base when it was dark then moved away from the light and back into the base. Additional preliminary tests performed under lighted conditions corroborated these results and showed that thrips avoid or move away from the light at 14 °C (see Appendix S2, Tables S1–S3). At 18 °C more thrips moved out of the base when it was dark than did so at 14 °C, and only a small percentage moved back into the base when the lights were turned on. At higher temperatures, thrips are attracted strongly to light as indicated by a significant movement towards the light, with almost all thrips moving into the arms at 26 °C. These higher temperatures are close to those at which cones typically begin their thermogenic activity.

Figure 3.

Percentage of Cycadothrips in Y-tube base (solid lines) and arms (dotted lines) at each uniform test temperature (14, 18, 22, and 26 °C), in dark-to-light changeover lighting experiments. Thrips' positions were first scored after 5 min of darkness (closed symbols) and then after 5 min of light (open symbols); that is, the results from each test are read vertically for each temperature on the x-axis, with two data points (base and arms) for the dark phase and two for the subsequent light phase. See Appendix S4, Table S5 and Fig. S2. Solid arrows show the direction of the change in the percentage of thrips in the base (increasing–back to base; decreasing–away from base) after the lights are turned on. Similarly, dashed arrows indicate the direction of the change in the percentage of thrips in the arms after lights are turned on. Significant differences between responses during the dark and light phase were determined by paired t-tests at each temperature, ***< 0·0001. There were only a few thrips in the Y-tube stem and the percentage there can be calculated as, % in stem = 100 - (% in base +% in arms).

Experiment 2: Heated base tests: do thrips move away from high temperatures?

Heated base experiments were performed in the dark to isolate the effect of temperature from the strong attractive effects of light just described. The results suggest that most thrips move away from temperatures ≥28 °C in the absence of any directional light stimulus (Fig. 4). At all Y-tube base temperatures greater than 22·3 °C, a significantly higher proportion of thrips moved out of the heated base than when the base was unheated (< 0·01, logistic regression models with temperature as a factor). Similarly, the proportion of thrips moving into the arms increased significantly when the base was heated to 28 and 36 °C above the ambient temperature (< 0·001, logistic regression model, except at 32 °C, = 0·048, because more thrips stayed in the Y-tube stem than moved into the arms at this temperature). No significant effects of gender (and no interactions with temperature) were detected in either the proportion remaining in the Y-tube base or moving into the arms (> 0·07). Nearly all thrips moved into the arms when the base temperature was 36 °C.

Figure 4.

The effects of heating the Y-tube base on the percentage of Cycadothrips that remain in the base and that move into the arms after 5 min in the dark. Y-tube arms were at room temperature of 22·3 ± 0·6 °C. The mean percentage in the Y-tube stem can be calculated by subtraction [% in stem = 100 − (% in base +% in arms)]. Means within the same line and marked with an asterisk are significantly different (*< 0·01, **< 0·001; ***< 0·0001, logistic regression of temperature of Y-tube base) from the mean recorded in the Y-tube base or arms at the base temperature of 22·3 °C.

Experiment 3: Y-tube heated arm tests: thrips response to thermogenic temperatures under different light conditions

With the ambient air and all of the Y-tube kept at ~22·3 °C except for one heated Y-tube arm, the percentage of thrips that moved from the Y-tube base to the heated arm was related to the heated arm's temperature as well as its light condition (Fig. 5 and Appendix S5, Table S6). First, with all parts of the Y-tube at 22·3 °C and in the dark, the thrips were almost evenly distributed in all Y-tube segments (Fig. 5, line 1), whereas thrips avoided that arm when it was heated to ≥ 28 °C, the same temperatures causing increased thrips departures from the base in heated base tests. Second, under conditions when the heated arm was darkened while the unheated arm was lighted, few thrips went to the heated arm at any temperature, while most went to the lighted, unheated arm (Fig. 5, line 2). Third, when all parts of the Y-tube were lighted (except the covered base), most thrips (84–95%) went into one or the other of the arms (Fig. 5, line 3) at 22·3 °C. There was a significant change in the Thrips' behaviour as the heated arm temperature increased to >32 °C, with fewer thrips moving into the heated arm at higher temperatures, and thrips avoiding the heated arm completely at 38 °C. Finally, when only the heated arm was lighted and all other segments were dark, the thrips were highly attracted to this arm until it reached ≥35 °C (Fig. 5, line 4). This result is mirrored in the straight tube temperature gradient test results (see Appendix S2, Fig. S1) which showed thrips starting to leave a lighted location at ~30 °C, and all having left only when the temperature reached ~ 45 °C. The results further illustrate the strong attractiveness of light at temperatures of ~22 °C and higher. Light induces an increased tolerance to heat at temperatures these thrips would avoid in the dark until the temperature is high enough to overcome the attractiveness of the light.

Figure 5.

Percentage of Cycadothrips that moved into the heated Y-tube arm (at indicated arm temperature, ±1 °C), after 5 min when: (i) all segments of the Y-tube were dark; (ii) the heated arm was dark and unheated arm (22·3 °C room temperature) was in the light; (iii) all segments were in the light; (iv) the heated arm was lighted, all other sections dark. Within a line, mean values associated with an asterisk are significantly different from the response to the ‘heated’ Y-tube arm at 22·3 °C (*< 0·01; **< 0·001). In line 2), there was no significant difference between the response at 22·3 °C and response at other temperatures.

Experiment 4: Thrips response to ß-myrcene at different ambient temperatures

The percentage of thrips that went to the β-myrcene treatment arm (of the total that went to both arms) under dark-to-light changeover conditions and at uniform temperatures of 14, 18, 22 or 26 °C depended on the concentration of β-myrcene (Fig. 6). Regardless of the temperature and lighting conditions, thrips were slightly attracted to, or neutral towards, the 5 μL β-myrcene arm relative to the control response (in the light the 5 μL β-myrcene was only slightly higher than controls = 0·017, but in the dark the response to 5 μL β-myrcene was significantly higher than the controls), but clearly avoided the two higher levels of β-myrcene (see Appendix S6, Table S7). As in the results from Experiment 1 when no β-myrcene was present (Fig. 3), the percentage of thrips that moved out of the base and into the arms was affected primarily by temperature and light, with a few exceptions (see Appendix S6, Figs. S3 and S4, and Tables S8 and S9). Most notably, in the light at 14 °C more thrips moved into the arms at the 200 μL β-myrcene levels than at the lower levels; and at 18 and 22 °C in the light, more thrips moved into the arms at both of the higher β-myrcene levels than did so at either the control or 5 μL β-myrcene level.

Figure 6.

Response of Cycadothrips to different concentrations of β-myrcene for different uniform temperature conditions in combination with different light conditions. Thrips' responses to all β-myrcene treatment levels were significantly different from the control, 0 β-myrcene (see Appendix S6, Table S7). The responses to the 200 μL β-myrcene levels are not shown but are similar and overlap those at the 100 μL level (<8·0% of thrips moved into the 200 μL β-myrcene arm).

Experiments 5 and 6: Thrips responses to carbon dioxide and relative humidity

There does not appear to be an effect of CO2, at the 2 and 5% levels, on thrips behaviour relative to the air control (see Appendix S7, Fig. S5). The responses to higher CO2 concentrations (up to 10 and 20%) vs. ambient air levels (under conditions of symmetric lighting only, ambient room temperature of ~24 °C) also indicate no significant effect of CO2 (Fig. 7, top half). Relative humidity did affect thrips behaviour, as significantly more thrips went to the dry air arm of the Y-tube in every comparison, regardless of the CO2 level (Fig. 7, lower half).

Figure 7.

Percentage of Cycadothrips that went to the test arm at different CO2 concentrations (dry (<8% RH) or humidified (~ 88% RH) vs. dry or humidified air; or humidified air vs. dry air). Tests were conducted at 24 °C in the light. The vertical line at 50% represents the neutral, or no preference response. **< 0·005, significant difference between% response between left treatment and right side treatment; all others are not significantly different (> 0·01), t-test. Total number of trials and average no. of thrips per trial is shown in parentheses (right hand column). On average for each comparison, >91% of thrips moved into the Y-tube arms.


This study extends our knowledge of how the Cycadothrips-Macrozamia pollination system functions by clarifying the ways in which C. chadwicki individuals respond to the multiple dynamically changing signals present during the cones' pollination phase, particularly during their metabolically intense thermogenic events. Increased cone temperatures, β-myrcene and relative humidity levels, and exterior light each motivate thrips to move but increased CO2 levels do not. Tables 1 and 2 summarize Thrips' responses to the signals tested and our translation of those responses into predicted Thrips' behaviour within cones.

Table 1. Summary of responses of Cycadothrips to different signals in Y-tube experiments; specified as, ‘Attract’ thrips, thrips ‘Avoid’ or ‘Not attractive’, or ‘No response’; ‘+’ = a high percentage responded; ‘−’ = a low percentage responded; and ‘NT’, not tested
Response toTemperature (°C)
Night, ~14–15Daytime, ~22–26Thermogenic, 28–40
  1. a

    Combinations of conditions not encountered in situ. High humidity may occur inside a cone during rain on a cold day.

DarkAttract+Not attractive+Not attractive+
ß-myrcene- low levelAttract-Slightly attract+NT, Slightly attract+
ß-myrcene- high (thermogenic) levelAvoid−aAvoid+aAvoid+
High humidityNTaAvoid+NT
High CO2NTaNo responseNT
Table 2. Expected behaviour of Cycadothrips within a cone, as translated from Table 1 summary, specified as, ‘Stay’ in cone or ‘Leave’ cone under each combination of conditions; ‘−’ = low percentage of thrips respond; ‘+’ = a high% response; ‘++’ = almost all thrips respond; and, ‘NT’, not tested
Response toCone temperature (°C)
Night, 14Day time, ~ 22–26Thermogenic, 28–40
  1. a

    Combinations of conditions not encountered in the field, except possibly high humidity inside the cone during rain on a cold day.

Light- outside coneStay+Leave+Leave++
Dark - both cone & ambientStay+Leave−Leave+
ß-myrcene - low in coneStay+LeaveLeave+ NT
ß-myrcene - high in coneStay?aLeave+Leave++ NT
High humidity in coneStay NTaLeave+Leave++ NT
High CO2 in coneStay NTaLeave+Leave++ NT

Temperature alone clearly affects thrips behaviour, with the higher temperatures typical of cone thermogenesis (>~28 °C) being unattractive, and indeed repellent, especially in the dark (Experiments 2 and 3). In comparison, the reduced movement of thrips at 14 °C (Experiment 1) likely reflects a direct effect of low temperature on the neuromuscular physiology of these small ectothermic animals. However, some thrips were able to move at 14 °C, so this is above their chill-coma temperature, where neuromuscular activity and motor control are minimal but with the effects reversible (Mutchmor 1967; Goller & Esch 1990). Temperature also affected Thrips' responses to light as seen by a reversal in their responses to light as temperature increases: avoidance at low temperatures and strong attraction at higher temperatures.

With respect to β-myrcene, at all temperatures tested thrips were slightly attracted, or neutral, to the low concentration tested, but avoided both higher concentrations tested, although Thrips' responses at colder temperatures were suppressed as reflected by only a small proportion of thrips moving into the arms (Experiment 4). These findings corroborate the push-pull, concentration-dependent, effect found at 24 °C (Terry et al. 2007b). Similar dose-dependent attractant to repellent reactions by thrips and other insects to an odourant have been reported (Davidson et al. 2008). The 200 μL level β-myrcene concentration tested is likely higher than what thrips experience inside of thermogenic cones (but not the 100 μL level, see Appendix S1); the 200 μL level was tested specifically to determine if very high levels would induce more thrips to move to the arms at the lower temperatures, which they did, but only in the light.

Thrips preferred dry to highly humidified air (Experiment 6), although their response was not as strong compared to temperature, light or β-myrcene. Insects, especially those that feed on mammalian hosts, are known to detect and respond positively to particular relative humidity values and temperatures (Altner & Loftus 1985; Lorenzo & Lazzari 1999; Barrozo, Manrique & Lazzari 2003; von Arx et al. 2012). Because thrips can easily succumb in water droplets, being able to detect and avoid high humidity associated with liquid water could be beneficial.

The lack of a behavioural response by thrips to CO2 (Experiments 5 and 6) was unexpected because most arthropod species that have been tested, including blood sucking insects, herbivores, detritivores, nest dwelling social insects and pollinators change behaviour in response to changing CO2 levels, with some floral structures reaching up to 10% CO2, as in some fig syconia (Galil, Zeroni & Bogoslavsky 1973; Nicolas & Sillans 1989; Goyret, Markwell & Raguso 2008; Guerenstein & Hildebrand 2008). Previous electro-physiological studies (Terry et al. 2007a) demonstrated that the antennal receptors of C. chadwicki do not respond to CO2. Because some insects have CO2 receptors on their labial palps or other mouthparts (Stange 1992; Guerenstein & Hildebrand 2008), we could not rule out the possibility of a behavioural response to CO2 before the current study.

Some of our behavioural tests were conducted under settings that present combinations of signals that thrips might not experience in nature (e.g. bright light within a warm cone, rather than such light outside of it; and thermogenic cone temperatures or high ß-myrcene concentrations in complete darkness). Such ‘conflicting signal tests’ enhance our understanding of the relative strengths of different signals, especially by showing the strong influence of light, which can attract thrips at temperatures above those that they would normally avoid. In addition, some test conditions could become applicable in the future, e.g., if climate change increases overnight temperatures to values that induce cone thermogenesis at night. Demonstrating Thrips' responses to such conditions sheds light on the potential robustness of the pollination system under different climate scenarios. Our data suggest that thrips would leave the cone's interior in the dark if thermogenesis occurred then, although we cannot say whether they would take flight and successfully move between cones.

Of all the signals tested, the only condition that appears to be necessary to initiate a mass movement of thrips is high temperature (Tables 1 and 2), and those temperatures interact with the other variables in complex ways. For example, a high level of β-myrcene or a high relative humidity may be sufficient to induce thrips mass departure from cones, but only in the presence of warmer temperatures. In addition, these two cone signals (which are associated with thermogenesis) are redundant with temperature, and all three are complementary with the ambient light signal. However, the response of thrips to light is also temperature-dependent – they avoid light at low temperatures and are attracted to light at high temperatures. This result is consistent with observations of Cycadothrips in the field. On cold days, thrips remain inside cold cones even when there is bright external light (Terry et al. 2004b) Furthermore, if thrips are shaken from cold cones, they crawl under nearby objects presumably to hide from the light. When shaken from warm cones, they typically take flight. Although temperature alone seems sufficient to drive thrips from cones under our test conditions, thrips may behave differently inside of cones where other rewards might be present, e.g., pollen or other nutrients.

Other Macrozamia species have different suites of cues during thermogenesis, for example high ß-myrcene levels do not occur during thermogenesis in M. communis cones (Wallenius et al. 2012). Our results suggest that pollinators might still leave thermogenic cones in such species if there is a sufficient rise in ambient/cone temperature and relative humidity. Only a few of the ~40 species of Macrozamia have had their cone traits investigated in detail (Mound & Terry 2001; Terry 2001; Seymour, Terry & Roemer 2004; Terry et al. 2004a,b, 2008; Roemer, Terry & Walter 2008; Roemer et al. 2012; Wallenius et al. 2012). Since some Macrozamia are pollinated by other species of Cycadothrips, or by Tranes weevils only, or by both types of insect, further studies are needed to investigate how cone traits vary and function relative to their specialist pollinator, including signals that attract these pollinators.

Several studies have confirmed that plant-pollinator, as well as plant-herbivore, interactions may involve more than one sensory modality of the insect (e.g., Raguso & Willis 2002; Angioy et al. 2004; Goyret, Markwell & Raguso 2007, 2008; Goyret et al. 2009). Our study adds to the understanding of such complex interactions because the results allow us to examine how different hypotheses of complex signal evolution, as outlined by Hebets & Papaj (2005), might apply to this cycad pollination system. This system involves multimodal complex signals that have duplicate functions, that is, all signals contribute towards expelling thrips from thermogenic cones. The redundant signal hypothesis within the ‘content-based signal category’ of hypotheses states that multiple signals convey the same information about the sender, for example the quality of the sender, and these signals increase the accuracy of the reception by the receiver. This hypothesis assumes a tight covariance among the signals and between the quality of the signals and signaller quality. These conditions appear to be true for the cycad system. Under the ‘efficacy-based category’, the efficacy back-up hypothesis contends that more signals help to yield a better response where there is environmental noise or variation that would tend to weaken any one signal. For these cycads, light levels, ambient temperatures or wind varies in different habitats and from day to day with different weather patterns, such that backup signals would increase the efficiency of the system. Finally, under the ‘inter-signal category’ of hypotheses, the context hypothesis might apply where the receiver responds to one signal in a context dependent manner. In the cycad system, thrips respond to abiotic and biotic signals only when cone temperatures are sufficiently high, which prevents thrips from leaving the cone in cold temperatures. While it is unlikely that very high ß-myrcene emissions would ever occur inside cold cones, high light levels do occur on cold days. Hebets & Papaj (2005) emphasize that different hypotheses may not be mutually exclusive. More importantly, to relate any particular selection process to these and other signals in this pollination system, we not only need a clear understanding of the signals that induce the pollinator to leave cones, but also of those involved in the attraction phase, which are part of our ongoing studies.

Finally, the proposed functions of thermogenesis in plants, and its associated benefits, range from protection of floral organs from freezing as in the skunk cabbage or aiding the development of these organs, to increasing the vapour pressure of odourants (Uemura et al. 1993), and/or to a direct effect on pollinator behaviour or larval development of insect pollinators. Many of the demonstrated functions involve pollinator activity or facilitation of the pollination process (Seymour, White & Gibernau 2003, 2009). Most of these functions are for thermogenesis acting alone or as a by-product of another signal, but few studies have tested for more than one signal, either a direct or indirect one. One notable exception is a study of the inflorescence of the dead-horse arum, Helicodiceros muscivorus, which demonstrated that thermogenesis increased the potential pollination effectiveness when both heating and odour were part of the inflorescence signal. A thermogenic appendix of the inflorescence creates a more realistic mimic of its pollinator's dead animal host by also emitting a foul rotting odour (Stensmyr et al. 2002; Angioy et al. 2004). In some species of Annonaceae and Araceae, thermogenesis is concurrent with both increased volatile emission and the arrival of pollinators (Skubatz et al. 1996; Gottsberger 1999; Gibernau & Barabé 2002; Seymour, White & Gibernau 2003; Kumano & Yamaoka 2006). Further study is needed to determine whether thermogenic temperatures provide a direct attractant or whether they serve some other function such as increasing the vapour pressure of the emitted volatiles. In our Macrozamia-Cycadothrips pollination system, thrips are leaving cones at the peak of thermogenesis, so while the dispersal of odours is likely enhanced by thermogenesis, the immediate effects of the combined signals on the pollinators is not for attraction but to motivate them to leave the cone. In short, despite the collective advances made in understanding thermogenesis in plants, we are still some way to a full functional interpretation of the process.


This study was supported in part by Australian Research Council's Discovery Project's funding scheme (project number DP1095482), and the University of Queensland. We thank Renee Rossini, Michelle Gleeson, Lucy Hurrey, and Michelle Rafter for field collection help, and the staff at Queensland Parks & Wildlife Service and Department of Environment & Heritage Protection that manage the reserves. We also thank Fred Adler, University of Utah, for statistical advice.